path: root/xc/doc/specs/Xserver/ddx.tbl.ms

.\" $XConsortium: ddx.tbl.ms,v 1.19 89/03/28 14:01:10 rws Exp $
.EF 'Porting Layer Definition'- % -'March 1, 1988'
.OF 'Porting Layer Definition'- % -'March 1, 1988'
.EH '''
.OH '''
.TL
Definition of the Porting Layer 
for the X v11 Sample Server
.AU
Susan Angebranndt
.AU
Raymond Drewry
.AU
Philip Karlton
.AU
Todd Newman
.AI
Digital Equipment Corporation
.sp
minor revisions by
.AU
Bob Scheifler
.AI
Massachusetts Institute of Technology

.LP
The following document explains the
structure of the X Window System display server and the interfaces among the larger pieces.
It is intended as a reference for programmers who are implementing an X Display Server
on their workstation hardware.
It is included with the X Window System source tape,
along with the document "Strategies for Porting the X v11 Sample Server."
The order in which you should read these documents is:

.IP 1) 
Read the first section 
of the "Strategies for Porting" document (Overview of Porting Process).

.IP 2) 
Skim over this document (the Definition document).

.IP 3) 
Skim over the remainder of the Strategies document.

.IP 4) 
Start planning and working, referring to the Strategies
and Definition documents.

You may also want to look at the following documents:
.IP \(bu 5
"The X Window System"
for an overview of X.
.IP \(bu 5
"Xlib - C Language X Interface"
for a view of what the client programmer sees.
.IP \(bu 5
"X Window System Protocol"
for a terse description of the byte stream protocol
between the client and server.
.IP \(bu 5
"X11 Server Extensions Engineering Specification"
for a description of how to add features to your X server
in an agreeable way.
.LP
UNIX is a trademark of AT&T.
QVSS, LK201, ULTRIX, VMS, DEC, MicroVAX and VAX are trademarks of Digital Equipment Corporation.
Macintosh and Apple are trademarks of Apple Computer, Inc.
PostScript is a trademark of Adobe Systems, Inc.
Ethernet is a trademark of Xerox Corporation.
The X Window System is a trademark of Massachusetts Institute of Technology.
Cray is a trademark of Cray Research, Inc.

.LP
To understand this document and the accompanying source
code, you should know the C language.
You should be familiar with 2D graphics and windowing
concepts such as clipping, bitmaps,
fonts, etc.
You should have a general knowledge of the X Window System.
To implement the server code on your hardware,
you need to know a lot about
your hardware, its graphic display device(s),
and (possibly) its networking and multitasking facilities.

This document depends a lot on the source code,
so you should have a listing of the code handy.
.LP
Some source on the distribution tape is directly compilable
on your machine.
Some of it will require
modification.
Other parts may have to be completely written from scratch.
.LP
The tape also includes source for a sample implementation of a display server on a
Digital Equipment Corporation QVSS workstation.  Although it has a monochrome
display, the sample server demonstrates many techniques that you will find useful
for implementing any type of X server.


.NH 1
The X Window System
.XS
The X Window System
.XE
.LP
The X Window System, or simply "X," is a
windowing system that provides high-performance, high-level,
device-independent graphics.

X is a windowing system designed for workstations with a lot of CPU power.
This means Digital's MicroVAX, Motorola's 68000, 
Intel's 80386, National Semiconductor's 32032 or a similarly powerful CPU with
1 Mbyte or more of main memory,
probably a hard disk,
and a display with with 
150,000 to 2 million  or more pixels on
the screen.
The workstation can have a simple, monochrome display or it can
have a color display with up to 32 bits per pixel with a special graphics processor
doing the work.
(In this document, monochrome means a black and white display with
one bit per pixel.
Even though the usual meaning of monochrome is more general, this special
case is so common that we decided to reserve the word for this purpose.)

X is designed for a networking environment where 
users can run applications on machines other than their own workstations.
Sometimes, the connection is over an Ethernet network with a protocol such as TCP/IP;
but, any "reliable" byte stream is allowable.
A high-bandwidth byte stream is preferable; RS-232 at
9600 baud would be slow.

X by itself allows great freedom of design.
For instance, it does not include any user interface standard.
Its intent is to "provide mechanism, not policy."
By making it general, it can be the foundation for a wide
variety of interactive software.

For a more detailed overview, see the document "The X Window System."
For details on the byte stream protocol, see "X Window System protocol."

.NH 1
OVERVIEW OF THE SERVER
.XS
OVERVIEW OF THE SERVER
.XE
.LP
The display server
manages windows and simple graphics requests
for the user on behalf of different client applications.
The client applications can be running on any machine on the network.
The server mainly does three things:
.IP \(bu 5
Responds to protocol requests from existing clients 
(mostly graphic and text drawing commands)
.IP \(bu 5
Sends device input (keystrokes and mouse actions) and other events to existing clients
.IP \(bu 5
Maintains client connections

.LP
The server code is organized into four major pieces:

.IP \(bu 5
Device Independent (DIX) layer - code 
shared among all implementations
.IP \(bu 5
Operating System (OS) layer - code 
that is different for each operating system
but is shared among all graphic 
devices for this operating system
.IP \(bu 5
Device Dependent (DDX) layer - code that is (potentially)
different for each combination of operating
system and graphic device
.IP \(bu 5
Extension Interface - a standard way to add
features to the X server

.LP
The "porting layer" consists of the OS and DDX layers; these are
actually parallel and neither one is on top of the other.
The DIX layer is intended to be portable 
without change to target systems and is not
detailed here, although several routines 
in DIX that are called by DDX are
documented.
Extensions incorporate new functionality into the server.
The interface to extensions is described in "X11 Server Extensions Engineering Specification."
.LP
The following sections outline the functions of the layers.
Section 2 briefly tells what you need to know about the DIX layer.
The OS layer is explained in Section 3.
Section 4 gives the theory of operation and procedural interface for the
DDX layer.

.NH 2
Notes On Resources and Large Structs
.XS
Notes On Resources and Large Structs
.XE
.LP
X resources are C structs inside the server.
Client applications create and manipulate these objects 
according to the rules of the X byte stream protocol.
Client applications refer to resources with resource IDs, 
which are 32-bit integers that are sent over the network.
Within the server, of course, they are just C structs, and we refer to them
by pointers.

The DDX layer has several kinds of resources:
.IP \(bu 5
Window 
.IP \(bu 5
Pixmap
.IP \(bu 5
Screen
.IP \(bu 5
Device
.IP \(bu 5
Colormap
.IP \(bu 5
Font
.IP \(bu 5
Cursor
.IP \(bu 5
Graphics Contexts
.LP
The type names of the more 
important server 
structs usually end in "Rec," such as "DeviceRec;"
the pointer types usually end in "Ptr," such as "DevicePtr."

The structs and
important defined constants are declared
in .h files that have names that suggest the name of the object.
For instance, there are two .h files for windows,
window.h and windowstr.h.
window.h defines only what needs to be defined in order to use windows 
without peeking inside of them;
windowstr.h defines the structs with all of their components in great detail
for those who need it.
.LP
Three kinds of fields are in these structs:
.IP \(bu 5
Attribute fields - struct fields that contain values like normal structs
.IP \(bu 5
Pointers to procedures that operate on the object
.IP \(bu 5
A private field (or two) used by your DDX code to keep private data
(probably a pointer
to another data structure)
.LP
DIX calls through
the struct's procedure pointers to do its tasks.
These procedures are set either directly or indirectly by DDX procedures.
Most of
the procedures described in the remainder of this
document are accessed through one of these structs.
For example, the procedure to create a pixmap
is attached to a ScreenRec and might be called by using the expression
.nf

        (* pScreen->CreatePixmap)(pScreen, width, height, depth).

.fi
All procedure pointers must be set to some routine unless noted otherwise;
a null pointer will have unfortunate consequences.

Procedure routines will be indicated in the documentation by this convention:
.nf

	void pScreen->MyScreenRoutine(arg, arg, ...)

.fi
as opposed to a free routine, not in a data structure:
.nf

	void MyFreeRoutine(arg, arg, ...)

.fi

The attribute fields are mostly set by DIX; DDX should not modify them 
unless noted otherwise.

.NH 1
DIX LAYER
.XS
DIX LAYER
.XE
.LP
The DIX layer is the machine and device independent part of X.
The source should be common to all operating systems and devices.
The port process should not include changes to this part, therefore internal interfaces to DIX 
modules are not discussed, except for public interfaces to the DDX and the OS layers.

In the process of getting your server to work, if
you think that DIX must be modified for purposes other than bug fixes,
you may be doing something wrong.
Keep looking for a more compatible solution.
When the next release of the X server code is available,
you should be able to just drop in the new DIX code and compile it.
If you change DIX,
you will have to remember what changes you made and will have
to change the new sources before you can update to the new version.

The heart of the DIX code is a loop called the dispatch loop.
Each time the processor goes around the loop, it sends off accumulated input events
from the input devices to the clients, and it processes requests from the clients.
This loop is the most organized way for the server to
process the asynchronous requests that
it needs to process.
Most of these operations are performed by OS and DDX routines that you must supply.

.NH 1
OS LAYER
.XS
OS LAYER
.XE
.LP
This part of the source consists of a few routines that you have to rewrite 
for each operating system.
These OS functions maintain the client connections and schedule work 
to be done for clients.  
They also provide an interface to font files,
font name to file name translation, and
low level memory management.

.nf
	void OsInit()
.fi
OsInit initializes your OS code, performing whatever tasks need to be done.
Frequently there is not much to be done.
The sample server implementation is in server/os/4.2bsd/osinit.c.

.NH 2
Scheduling and Request Delivery
.XS
Scheduling and Request Delivery
.XE
.LP
The main dispatch loop in DIX creates the illusion of multitasking between 
different windows, while the server is itself but a single process.
The dispatch loop breaks up the work for each client into small digestible parts.
Some parts are requests from a client, such as individual graphic commands.
Some parts are events delivered to the client, such as keystrokes from the user.
The processing of events and requests for different
clients can be interleaved with one another so true multitasking
is not needed in the server.

You must supply some of the pieces for proper scheduling between clients.
.nf

	int WaitForSomething(pClientReady)
		int *pClientReady;
.fi
.LP
WaitForSomething is the scheduler procedure you must write that will
suspend your server process until something needs to be done.   
This call should
make the server suspend until one or more of the following occurs:
.IP \(bu 5
There is an input event from the user or hardware (see SetInputCheck())
.IP \(bu 5
There are requests waiting from known clients, in which case 
you should return a count of clients stored in pClientReady
.IP \(bu 5
A new client tries to connect, in which case you should create the
client and  then continue waiting
.LP
If WaitForSomething() decides it is about to do something that might block
(in the sample server,  before it calls select()) it must call a DIX
routine called BlockHandler().
.nf

	BlockHandler(pTimeout, pReadmask)
		pointer pTimeout;
		pointer pReadmask;
.fi
The types of the arguments are for agreement between the OS and DDX
implementations,  but the pTimeout is a pointer to the information
determining how long the block is allowed to last,  and the
pReadmask is a pointer to the information describing the descriptors
that will be waited on.
.LP
In the 4.2 case,  pTimeout is a struct timeval **,  and pReadmask is
the address of the select mask for reading.
.LP
Immediately after it returns from the
block,  even if it didn't actually block,  it must call the DIX routine
WakeupHandler().
.nf

	WakeupHandler(result, pReadmask)
		unsigned long result;
		pointer pReadmask;
.fi
.LP
Once again,  the types are not specified by DIX.  The result is the
success indicator for the thing that (may have) blocked,
and the pReadmask is a mask of the descriptors that came active.
.LP
In the 4.2 case,  result is the resdult from select(),  and pReadmask is
the address of the select mask for reading.
.LP
The DIX BlockHandler() iterates through the Screens,  for each one calling
its BlockHandler.  A BlockHandler is declared thus:
.nf

	void xxxBlockHandler(nscreen, pbdata, pptv, pReadmask)
		int nscreen;
		pointer pbdata;
		pointer pptv;
		pointer pReadmask;
.fi
The arguments are the index of the Screen,  the BlockData field
of the Screen,  and the arguments to the DIX BlockHandler().
.LP
The DIX WakeupHandler() does the same thing,  calling each Screen's
WakeupHandler.  A WakeupHandler is declared thus:
.nf

	void xxxWakeupHandler(nscreen, pbdata, err, pReadmask)
		int nscreen;
		pointer pbdata;
		unsigned long err;
		pointer pReadmask;
.fi
The arguments are the index of the Screen,  the BlockData field
of the Screen,  and the arguments to the DIX BlockHandler().
.LP
The WaitForSomething on the sample server also has a built
in screen saver that darkens the screen if no input happens for a period of time.
The sample server implementation is in server/os/4.2bsd/WaitFor.c.
.LP
Note that WaitForSomething() may be called when you already have several
outstanding things (events, requests, or new clients) queued up.
For instance, your server may have just done a large graphics request,
and it may have been a long time since WaitForSomething() was last called.
If many clients have lots of requests queued up, DIX will only service
some of them for a given client
before going on to the next client (see isItTimeToYield, below).
Therefore, WaitForSomething() will have to report that these same clients
still have requests queued up the next time around.
.LP
An implementation should return information on as
many outstanding things as it can.
For instance, if your implementation always checks for client data first and does not
report any input events until there is no client data left,
your mouse and keyboard might get locked out by an application that constantly
barrages the server with graphics drawing requests.
.LP
A list of indexes (client->index) for clients with data ready to be read or
processed should be returned in pClientReady, and the count of indexes
returned as the result value of the call.
This is not clients that have full requests ready, but any clients who have
any data ready to be read or processed.
The DIX dispatcher
will process requests from each client in turn by calling 
ReadRequestFromClient(), below.   
.LP
WaitForSomething() must create new clients as they are requested (by
whatever mechanism at the transport level).  A new client is created
by calling the DIX routine:
.nf

	ClientPtr NextAvailableClient(ospriv)
		pointer ospriv;
.fi
This routine returns NULL if a new client cannot be allocated (e.g. maximum
number of clients reached).  The ospriv argument will be stored into the OS
private field (pClient->osPrivate), to store OS private information about the 
client.  In the sample server, the osPrivate field contains the 
number of the socket for this client. See also "New Client Connections."
NextAvailableClient() will call InsertFakeRequest(), so you must be
prepared for this.
.LP
If there are outstanding input events,
you should make sure that the two SetInputCheck() locations are unequal.
The DIX dispatcher will call your implementation of ProcessInputEvents()
until the SetInputCheck() locations are equal.
.LP
The sample server contains an implementation of WaitForSomething()
that is portable to UNIX 4.2 systems and to other systems as well.
In it, the
following two routines indicate to WaitForSomething() what devices should
be waited for.   FID is an OS dependent type; in the sample server
it is an open file descriptor.
.nf

	void AddEnabledDevice(fd)
		FID fd;

	void RemoveEnabledDevice(fd)
		FID fd;
.fi
These two routines are
called from the initialize cases of the
Input Procedures that are stored in the DeviceRec (the
routine passed to AddInputDevice()).
The sample server implementation is in server/os/4.2bsd/connection.c.
.nf

	Bool isItTimeToYield;
.fi
.LP
isItTimeToYield is a global variable you can set 
if you want to tell
DIX to end the client's "time slice" and start paying attention to the next client.
After the current request is finished, DIX will move to the next client.
.LP
In the sample
server, ReadRequestFromClient() sets isItTimeToYield after
10 requests packets in a row are read from the same client.
.LP
This scheduling algorithm can have a serious effect upon performance when two
clients are drawing into their windows simultaneously.
If it allows one client to run until its request 
queue is empty by ignoring isItTimeToYield, the client's queue may
in fact never empty and other clients will be blocked out.
On the other hand, if it switchs between different clients too quickly,
performance may suffer due to too much switching between contexts.
For example, if a graphics processor needs to be set up with drawing modes
before drawing, and two different clients are drawing with
different modes into two different windows, you may 
switch your graphics processor modes so often that performance is impacted.
.LP
See the Strategies document for 
heuristics on setting isItTimeToYield.

.NH 2
New Client Connections
.XS
New Client Connections
.XE
.LP
The process whereby a new client-server connection starts up is 
very dependent upon what your byte stream mechanism.
This section describes byte stream initiation using examples from the TCP/IP
implementation on the sample server.
.LP
The first thing that happens is a client initiates a connection with the server.
How a client knows to do this depends upon your network facilities and the
Xlib implementation.
In a typical scenario, a user named Fred 
on his X workstation is logged onto a Cray
supercomputer running a UNIX shell in an X window.  Fred can type shell
commands and have the Cray respond as though the X server were a dumb terminal.
Fred types in a command to run an X client application that was linked with Xlib.
Xlib looks at the UNIX environment variable DISPLAY, which has the 
value "fredsbittube:0.0."
The host name of Fred's workstation is "fredsbittube," and the 0s are 
for multiple screens and multiple X server processes.
(Precisely what 
happens on your system depends upon how X and Xlib are implemented.)
.LP
The client application calls a TCP routine on the 
Cray to open a TCP connection for X
to communicate with the network node "fredsbittube."
The TCP software on the Cray does this by looking up the TCP
address of "fredsbittube" and sending an open request to TCP port 6000
on fredsbittube.  
.LP
All X servers on TCP listen for new clients on port 6000;
this is known as a "well-known port" in IP terminology.
.LP
The server receives this request from its port 6000
and checks where it came from to see if it is on the server's list
of "trustworthy" hosts to talk to.
Then, it opens another port for communications with the client.
This is the byte stream that all X communications will go over.
.LP
Actually, it is a bit more complicated than that.
Each X server process running on the host machine is called a "display."
Each display can have more than one screen that it manages.
"corporatehydra:3.2" represents screen 2 on display 3 on 
the multi-screened network node corporatehydra.
The open request would be sent on well-known port number 6003.
.LP
Once the byte stream is set up, what goes on does not depend very much
upon whether or not it is TCP.
The client sends an xConnClientPrefix struct (see Xproto.h) that has the
version numbers for the version of Xlib it is running, some byte-ordering information, 
and two character strings used for authorization.
If the server does not like the authorization strings
or the version numbers do not match within the rules,
or if anything else is wrong, it sends a failure 
response with a reason string.
.LP
If the information never comes, or comes much too slowly, the connection will
should be broken off.  You must implement the connection timeout.  The
sample server implements this by keeping a timestamp for each still-connecting
client and, each time just before it attempts to accept new connections, it
closes any connection that are too old.
The connection timeout can be set from the command line.
.LP
You must implement whatever authorization schemes you want to support.
The sample server on the distribution tape supports a simple authorization
scheme.  The only interface seen by DIX is:
.nf

	char * 
	ClientAuthorized(client, proto_n, auth_proto, string_n, auth_string)
	    ClientPtr client;
	    char *auth_proto, *auth_string;
	    int proto_n, string_n;
.fi
.LP
DIX will only call this once per client, once it has read the full initial
connection data from the client.  If the connection should be
accepted ClientAuthorized() should return NULL, and otherwise should
return an error message string.
.LP
Accepting new connections happens internally to WaitForSomething().
WaitForSomething() must call the DIX routine NextAvailableClient()
to create a client object.
Processing of the initial connection data will be handled by DIX.
Your OS layer must be able to map from a client
to whatever information your OS code needs to communicate
on the given byte stream to the client.
DIX uses this ClientPtr to refer to
the client from now on.   The sample server uses the osPrivate field in
the ClientPtr to store the file descriptor for the socket, the
input and output buffers, and authorization information.
.LP
To initialize the methods you choose to allow clients to connect to
your server, main() calls the routine
.nf

	CreateWellKnownSockets()
.fi
.LP
This routine is called only once, and not called when the server
is reset.  To recreate any sockets during server resets, the following
routine is called from the main loop:
.nf

	ResetWellKnownSockets()
.fi
Sample implementations of both of these routines are found in 
server/os/4.2bsd/connection.c.
.LP
For more details, see the section called "Connection Setup" in the X protocol specification.

.NH 2
Reading Data from Clients
.XS
Reading Data from Clients
.XE
.LP
Requests from the client are read in as a byte stream by the OS layer.
They may be in the form of several blocks of bytes delivered in sequence; requests may
be broken up over block boundaries or there may be many requests per block.
Each request carries with it length information.
It is the responsibility of the following routine to break it up into request blocks.
.nf

	char *ReadRequestFromClient(who, status, oldpointer)
		ClientPtr who;
		int *status;
		char *oldpointer;
.fi
.LP
You must write
the routine ReadRequestFromClient() to get one request from the byte stream
belonging to client "who."
You must swap the third and fourth bytes (the second 16-bit word) according to the 
byte-swap rules of
the protocol to determine the length of the
request.  
This length is measured in 32-bit words, not in bytes.  Therefore, the 
theoretical maximum request is 256K.
(However, the maximum length allowed is dependent upon the server's input
buffer.  This size is sent to the client upon connection.  The maximum 
size is the constant MAX_REQUEST_SIZE in server/include/os.h)
The rest of the request you return is
assumed NOT to be correctly swapped for internal 
use, because that is the responsibility of DIX.
.LP
'who' is the ClientPtr returned from WaitForSomething.
The status should be set to the (positive) byte count if the read is successful, 
0 if the read was blocked, or a negative error code if an error happened.
.LP
You must then return a pointer to
the bytes of the request.
This can simply be a pointer into your buffer;
DIX may modify it in place but will not otherwise cause damage.
Of course, the request must be contiguous; you must 
shuffle it around in your buffers if not.
.LP
oldpointer is a pointer that was returned by you on a previous
call to ReadRequestFromClient() (probably the last).
You can consider the memory it points to
to be free buffer space that DIX no longer needs; you can put another request there.
The sample server, however, ignores this.

The sample server implementation is in server/os/4.2bsd/io.c.

.XS
Inserting Data for Clients
.XE
.LP
DIX can insert data into the client stream, and can cause a "replay" of
the current request.
.nf

	Bool InsertFakeRequest(client, data, count)
	    ClientPtr client;
	    char *data;
	    int count;

	ResetCurrentRequest(client)
	    ClientPtr client;
.fi
.LP
InsertFakeRequest() must insert the specified number of bytes of data
into the head of the input buffer for the client.  This may be a
complete request, or it might be a partial request.  For example,
NextAvailableCient() will insert a partial request in order to read
the initial connection data sent by the client.  The routine returns FALSE
if memory could not be allocated.  ResetCurrentRequest()
should "back up" the input buffer so that the currently executing request
will be reexecuted.  DIX may have altered some values (e.g. the overall
request length), so you must recheck to see if you still have a complete
request.  ResetCurrentRequest() should always cause a yield (isItTimeToYield).

.NH 2
Sending Events, Errors And Replies To Clients
.XS
Sending Events, Errors And Replies To Clients
.XE
.LP
.nf

	int WriteToClient(who, n, buf)
		ClientPtr who;
		int n;
		char *buf;
.fi
WriteToClient should write n bytes starting at buf to the 
ClientPtr "who".
It returns the number of bytes written, but for simplicity,
the number returned must be either the same value as the number
requested, or -1, signaling an error.
The sample server implementation is in server/os/4.2bsd/io.c.
.LP
.nf

	void FlushAllOutput()

	void FlushIfCriticalOutputPending()

	void SetCriticalOutputPending()
.fi
These three routines may be implemented to support buffered or delayed
writes to clients, but at the very least, the stubs must exist.
FlushAllOutput() unconditionally flushes all output to clients;
FlushIfCriticalOutputPending() flushes output only if
SetCriticalOutputPending() has be called since the last time output
was flushed.
The sample server implementation is in server/os/4.2bsd/io.c and
actually ignores requests to flush output on a per-client basis
if it knows that there
are requests in that client's input queue.
.NH 2
Font Support
.XS
Font Support
.XE
.LP
In the sample server, fonts are encoded in disk files.
There is one file per font, with a file name like "Fixed.snf," in 
one of a handful of directories.
All of these directories are listed in the current "font path."

In principle, you can put all your fonts in ROM or in RAM in your server.
You can put them all in one library file on disk.
You could generate them on the fly from stroke descriptions.
You can put them
anywhere you want as long as your scheme follows the rules described 
in this section.

The sample server implementation is in fileio.c and filenames.c.

.NH 3
Font Support Data Structures
.XS
Font Support Data Structures
.XE
.LP
A font is identified by a font name, which is a counted character string.
A font name is sent over the network and its length is 
kept as a separate integer (and it is NOT null terminated).
X assigns no format rules for font names, although 
some groups of fonts follow conventions.
Typical font names are "Fixed," "Times 12," "Avant Garde Demi 24."
.LP
Given a font name, your OS code must generate
a "font file name" for that font.
You will probably return a real filename for your operating system,
but anything that serves the same purpose will do, including the font name itself.
.LP
Given a font filename, you must be able to produce
font data in Server Natural Format (see the section on Fonts).
DIX thinks it is reading a file when it asks for the font data, and it calls
file-read-like routines to read it into its font buffers.
Probably these will be real files, but once again, you can do anything you want.
(If you want to use font data in a format other than SNF, you will have
to rewrite part of DIX.)
.LP
The font-name-to-font-filename translation procedure may want to 
use a font path to look up the font files.
(A path is a list of directories to search, in order.)
Not all servers need to implement a font path.
In fact, these functions can be very OS-specific and nonportable.
.LP
There is a struct, called a FontPathRec, that holds a list of character strings.
You use it to communicate the current path to and from DIX.
This struct is also used for other purposes, such as providing a list of 
font names that matches a given pattern.

A FontPathRec has a count of the number of strings it holds, a pointer
to an array of lengths, and a pointer to an array of char * pointers.
All of these arrays and strings are dynamically allocated.  The strings
are not NULL-terminated.
.nf

	void FreeFontRecord(pFP)
		FontPathPtr pFP;
.fi
FreeFontRecord must free dynamically allocated FontPathRec's and all underlying
structures.

When a GC is created, it is assigned default attributes such as solid fill style,
etc.  The default font is set in DIX to the font whose name is a defined
constant COMPILEDDEFAULTFONT, in the file server/include/site.h.
In the sample server, it is "fixed," a simple fixed-width font 
that is supplied as one of the fonts in bdf format.

.NH 3
Font Names and Filenames
.XS
Font Names and Filenames
.XE
.LP
.nf

	int ExpandFontName( ppFilename, lenFontname, pFontname)
		char **ppFilename;
		int lenFontname;
		char *pFontname;
.fi
This routine translates a font name into the corresponding filename.
pFontname points to the bytes of the font name, and lenFontname is the
length - pFontname is NOT null terminated.
ppFilename is a pointer to a char * variable belonging to the caller.

.nf

	FontPathPtr
	ExpandFontNamePattern(lenPattern, pPattern, maxNames)
		int lenPattern;
		char *pPattern;
		int maxNames;
.fi
This routine must take a pattern for a font name, 
search for all of the available fonts
that match that pattern, and return them in a pointer to a FontPathRec.
pPattern and lenPattern are the pattern string in typical "font name" format
(not null terminated).
maxNames is the maximum number of names to return; you can allocate
a FontPathRec of the appropriate size knowing this.

The pattern is case insensitive.  ? and * work 
as standard match characters (?=any character,
*=any sequence of characters).

.NH 3
Font Files
.XS
Font Files
.XE
.LP
.nf

	FID FiOpenForRead( lenFilename, filename)
		int lenFilename;
		char *name;

	int FiRead( buf, itemsize, nitems, fid)
		char *buf;
		unsigned itemsize;
		unsigned nitems;
		FID fid;

	int FiClose(fid)
		FID fid;
.fi
These routines are analogous to DDX SNF runtime library routines.
FID is an OS specific type.  
The open routine takes a font filename, as returned by ExpandFontName().
In other words, if font filenames really are not filenames, then these routines
should not open real files.
You must supply them if you use the DDX SNF library.
Refer to a C runtime library reference manual for more details.
.nf

	int SetFontPath( client, npaths, countedStrings)
		ClientPtr client;
		int npaths;
		char *countedStrings;
.fi
SetFontPath must set the entire path to the new path passed in.
client is the client issuing the request;
npaths is the number of paths in the list.
countedStrings is a pointer to a buffer that contains counted strings
for each directory name in the path; each counted string is a
length byte followed by that many characters (not padded, not null-terminated).
If npaths = 0, SetFontPath should reinstate the default font path.
The routine returns Success if the path is installed, else an error such
as BadAlloc.
.LP
The default font path is set in DIX to the string that is a defined
constant COMPILEDDEFAULTFONTPATH, in the file site.h.
In the sample server, it is "/usr/src/x11/fonts/snf/."
.nf

	FontPathPtr GetFontPath()
.fi
GetFontPath() must return the font path.
You must copy it into a dynamically allocated structure
that will eventually be freed with FreeFontRecord().
Servers that choose to not implement these should return an 
empty path.
.nf

	int SetDefaultFontPath(dirname)
		char *dirname;
.fi
You should write this routine to set the default font path.
DIX will pass you the value of COMPILEDDEFAULTFONTPATH when the 
server starts up.
Return Success on success, else an error such as BadAlloc.

.NH 2
Memory Management
.XS
Memory Management
.XE
.LP
Memory management is the same as in the UNIX runtime library.
Xalloc(), Xrealloc(),  and Xfree() work just like malloc(), 
realloc(), and free(),
except that you can pass a null pointer to Xrealloc() to have it allocate
anew or
pass a null pointer to Xfree() and nothing will happen.
The versions in the sample server also do some checking that is useful for debugging.
Consult a C runtime library reference manual for more details.

The macros ALLOCATE_LOCAL and DEALLOCATE_LOCAL are provided in
server/include/os.h.  These are useful if
your compiler supports alloca() (or some
method of allocating memory from the stack).  The sample server
just calls Xalloc() and Xfree().

Treat memory allocation carefully in your implementation.
Memory leaks can be very hard to find and are frustrating
to a user.  An X server could be running
for days or weeks without being reset, just like a regular terminal.
If you leak a few dozen k per day, that will add up and will cause problems
for users that leave their workstations on.

.NH 2
Other OS Functions
.XS
Other OS Functions
.XE
.LP
.nf
	void
	ErrorF(f, s0, s1, s2, s3, s4, s5, s6, s7, s8, s9)
	    char *f;
	    char *s0, *s1, *s2, *s3, *s4, *s5, *s6, *s7, *s8, *s9;

	void
	FatalError(f, s0, s1, s2, s3, s4, s5, s6, s7, s8, s9)
	    char *f;
	    char *s0, *s1, *s2, *s3, *s4, *s5, *s6, *s7, *s8, *s9;

	void
	Error(str)
	    char *str;
.fi
.LP
You should write
these three routines to provide for diagnostic output from the dix and
ddx layers, although implementing them to produce no output will not
affect the correctness of your server.
ErrorF() and FatalError() take
a printf() type of format specification in the first argument and
up to ten format arguments following that.
Normally, the formats passed to ErrorF() and FatalError() should be
terminated with a newline.
Error() provides an os interface for printing out the string passed
as an argument followed by a meaningful explanation of the last
system error.
Normally the string does not contain a newline, and it is only called
by the ddx layer.
In the sample implementation, Error() uses the Unix routine perror().
.LP
After printing the message arguments, FatalError() must be implemented
such that the server will call AbortDDX() to give the ddx layer
a chance to reset the hardware, and then
terminate the server; it must not return.
.LP
The sample server implementation for these routines
is in server/os/4.2bsd/util.c.

.NH 1
DDX LAYER
.XS
DDX LAYER
.XE
.LP
This section describes the
interface between DIX and DDX.
While there may be an OS-dependent driver interface between DDX
and the physical device, that interface is left to the DDX
implementor and is not specified here.
.LP
The DDX layer does most of its work through procedures that are
pointed to by different structs.
As previously described, the behavior of these resources is largely determined by
these procedure pointers.
Most of these routines are for graphic display on the screen or support functions thereof.
The rest are for user input from input devices.

.NH 2
INPUT
.XS
INPUT
.XE
.LP
In this document "input" refers to input from the user, 
such as mouse, keyboard, and
bar code readers.
X input devices are of three kinds: keyboard, pointing device, and
"other."
Other is
used for extensions; there is no support for other devices in the
core X server.
In fact, there is no support for other keyboards or 
pointing devices is available
beyond the one keyboard and pointing device that you register during initialization.
See the Extension document for more details.

You, the DDX programmer, are
responsible for some of the routines in this section.
Others are DIX routines that you should call to do the things you need to do in these DDX routines.
Pay attention to which is which.

.NH 3
Input Device Data Structures
.XS
Input Device Data Structures
.XE
.LP
DIX keeps a global directory of devices in a central data structure
called InputInfo.
For each device there is a device structure called a DeviceRec.
DIX can locate any DeviceRec through InputInfo.
In addition, it has a special pointer to identify the main pointing device
and a special pointer to identify the main keyboard.
.LP
The DeviceRec (server/include/input.h) is a device-independent
structure that contains the state of an input device.
A DevicePtr is simply a pointer to a DeviceRec.
.LP
An xEvent describes an event the server reports to a client.
Defined in Xproto.h, it is a huge struct of union of structs that have fields for
all kinds of events.
All of the variants overlap, so that the struct is actually very small in memory.

.NH 3
Processing Events
.XS
Processing Events
.XE
.LP
The main DDX input interface is the following routine:
.nf

	void ProcessInputEvents()
.fi
You must write this routine to deliver input events from the user.
DIX calls it when input is pending, and possibly 
even when it is not.  
You should write it to get events from each device and deliver
the events to DIX.
To deliver the events to DIX, you should call the following
routine:
.nf

	void DevicePtr->processInputProc(pEvent, device)
.fi
This is the "input proc" for the device, a DIX procedure.
DIX will fill in this procedure pointer to one of its own routines by 
the time ProcessInputEvents() is called the first time.
Call this input proc routine as many times as needed to
deliver as many events as should be delivered.
DIX will buffer them up and send them out as needed.

For example, your ProcessInputEvents() routine might check the mouse and the
keyboard.
If the keyboard had several keystrokes queued up, it could just call
the keyboard's processInputProc as many times as needed to flush its internal queue.

event is an xEvent struct you pass to the input proc.
When the input proc returns, it is finished with the event rec, and you can fill
in new values and call the input proc again with it.

device is a DevicePtr.

You should deliver the events in the same order that they were generated.

For keyboard and pointing devices the xEvent variant should be keyButtonPointer.
Fill in the following fields in the xEvent record:
.nf

	type		is one of the following: KeyPress, KeyRelease, ButtonPress, 
					ButtonRelease, or MotionNotify
	detail		for KeyPress or KeyRelease fields, this should be the 
					key number (not the ASCII code); otherwise unused
	time		is the time that the event happened (32-bits, in milliseconds, arbitrary origin)
	rootX		is the x coordinate of cursor
	rootY		is the y coordinate of cursor

.fi
The rest of the fields are filled in by DIX.
.LP
The time stamp is maintained by your code in the DDX layer, and it is your responsibility to 
stamp all events correctly.
.LP
The x and y coordinates of the pointing device and the time must be filled in for all event types
including keyboard events.
.LP
The pointing device must report all button press and release events.
In addition, it should report a MotionNotify event every time it gets called 
if the pointing device has moved since the last notify.
Intermediate pointing device moves are stored in a special GetMotionEvents buffer,
because most client programs are not interested in them.

The sample server implementation is in server/ddx/dec/qvss/qvss_io.c.

.NH 3
Telling DIX When Input is Pending
.XS
Telling DIX When Input is Pending
.XE
.LP
In the server's dispatch loop, DIX checks to see
if there is any device input pending whenever WaitForSomething() returns.  
If the check says that input is pending, DIX calls the
DDX routine ProcessInputEvents().
.LP
This check for pending input must be very quick; a procedure call
is too slow.
The code that does the check is a hardwired IF 
statement in DIX code that simply compares the values
pointed to by two pointers.
If the values are different, then it assumes that input is pending and
ProcessInputEvents() is called by DIX.
.LP
You must pass pointers to DIX to tell it what values to compare.
The following procedure
is used to set these pointers:
.nf

	void SetInputCheck(p1, p2)
		long *p1, *p2;
.fi
.LP
You should call it sometime during initialization to indicate to DIX the
correct locations to check.
You should 
pay special attention to the size of what they actually point to, 
because the locations are assumed to be longs.

These two pointers are initialized by DIX
to point to arbitrary values that
are different.
In other words, if you forget to call this routine during initialization,
the worst thing that will happen is that
ProcessInputEvents will be called when 
there are no events to process.

p1 and p2 might
point at the head and tail of some shared
memory queue. 
Another use would be to have one point at a constant 0, with the
other pointing at some mask containing 1s
for each input device that has
something pending.

The DDX layer of the sample server calls SetInputCheck()
once when the
server's private internal queue is initialized 
(server/ddx/dec/qvss/qvss_io.c, qvssMouseProc).
It passes pointers to the queue's head and tail.

.nf
	long TimeSinceLastInputEvent()
.fi
DDX must time stamp all hardware input
events.  But DIX sometimes needs to know the
time and the OS layer needs to know the time since the last hardware
input event in
order for the screen saver to work.   TimeSinceLastEvent() returns
the this time in milliseconds.  (See server/ddx/dec/qvss/qvss_io.c)

.NH 3
Controlling Input Devices
.XS
Controlling Input Devices
.XE
.LP
You must write four routines to do various device-specific 
things with the keyboard and pointing device.
They can have any name you wish because 
you pass the procedure pointers to DIX routines.

The sample server implementations are in server/ddx/dec/qvss/qvss_io.c.

.nf

	int pInternalDevice->GetMotionEvents(buff, start, stop)
		xTimecoord *buff;
		CARD32 start, stop;
.fi
You write this DDX routine to fill in buff with all the motion events that
have times (32-bit count of
milliseconds) between time start and time stop.
It should return the number of motion events returned.
If there is no motion events support, this routine should do nothing and return zero.
The maximum size is set in RegisterPointerDevice(), below.

When the user drags the pointing device,
the cursor position theoretically sweeps through an infinite
number of points.
Normally, a client that is concerned with points other than the starting and ending points
will receive a pointer-move event only as often as the 
server generates them. (Move events
do not queue up; each new one replaces the last in the queue.)
A server, if desired, can implement a scheme to save these intermediate events
in a motion buffer.
A client application, like a paint program, may then request that 
these events be delivered to it through this routine.
.nf

	void pInternalDevice->Bell(loud, pDevice)
		int loud;
		DevicePtr pDevice;
.fi
You need to write this routine to ring the bell on the keyboard. 
loud is a number from 0 to 100, with 100 being the loudest.
.nf

	void pInternalDevice->ControlProc(device, ctrl)
		DevicePtr device;
		SomethingCtrl *ctrl;

.fi
.LP
You write two versions of this procedure, one for the keyboard and one for the pointing device.
DIX calls it to inform DDX when a client has requested changes in the current
settings for the particular device.
For a keyboard, this might be the repeat threshold and rate.
For a pointing device, this might be a scaling factor (coarse or fine) for position reporting.
See input.h for the ctrl structures.

.NH 3
Input Initialization
.XS
Input Initialization
.XE
.LP
Input initialization is a bit complicated.
It all starts with InitInput(), a routine that you write to call 
AddInputDevice() twice
(once for pointing device and once for keyboard.)
You also want to call RegisterKeyboardDevice() and RegisterPointerDevice()
on them.

When you Add the devices, a routine you supply for each device
gets called to initialize them.
Your individual initialize routines must call InitKeyboardDeviceStruct()
or InitPointerDeviceStruct(), depending upon which it is.
In other words, you indicate twice that the keyboard is the keyboard and
the pointer is the pointer.
.nf

	void InitInput(argc, argv)
	    int argc;
	    char **argv;
.fi
.LP
InitInput is a DDX routine you must write to initialize the 
input subsystem in DDX.
It must call AddInputDevice() for each device that might generate events.
In addition, you must register the main keyboard and pointing devices by
calling RegisterPointerDevice() and RegisterKeyboardDevice().
The sample server implementation is in server/ddx/dec/qvss/init.c.
.nf

	DevicePtr AddInputDevice(deviceProc, autoStart)
		DeviceProc deviceProc;
		Bool autoStart;
.fi
.LP
AddInputDevice is a DIX routine you call to create a device object.
deviceProc is a DDX routine that is called by DIX to do various operations.
AutoStart should be TRUE for devices that need to be turned on at
initialization time with a special call, as opposed to waiting for some 
client application to
turn them on.
This routine returns NULL if sufficient memory cannot be allocated to
install the device.

Note also that except for the main keyboard and pointing device, 
an extension is needed to provide for a client interface to a device.
.nf

	void RegisterPointerDevice(device, numMotionEvents)
		DevicePtr device;
		int numMotionEvents;
.fi
.LP
RegisterPointerDevice is a DIX routine that your DDX code calls that
makes that device the main pointing device.  
This routine is called once upon initialization and cannot be called again.

numMotionEvents is for the motion-buffer-size for the GetMotionEvents
request.
A typical length for a motion buffer would be 100 events.
A server that does not implement this capability should set 
numMotionEvents to zero.
.nf

	void RegisterKeyboardDevice(device)
		DevicePtr device;
.fi
.LP
RegisterKeyboardDevice makes the given device the main keyboard.
This routine is called once upon initialization and cannot be called again.

The following DIX
procedures return the specified DevicePtr. They may or may not be useful
to DDX implementors.
.nf

	DevicePtr LookupKeyboardDevice()
.fi
.LP
LookupKeyboardDevice returns pointer for current main keyboard device.
.nf

	DevicePtr LookupPointerDevice()
.fi
.LP
LookupPointerDevice returns pointer for current main pointing device.

.LP
A DeviceProc (the kind passed to AddInputDevice()) in the following form:
.nf

	Bool pInternalDevice->DeviceProc(device, action);
		DevicePtr device;
		int action;
.fi
.LP
You must write a DeviceProc for each device.
device points to the device record.
action tells what action to take;
it will be one of  these defined constants  (defined in input.h):
.IP \(bu 5
DEVICE_INIT -
At DEVICE_INIT time, the device should initialize itself by calling
InitPointerDeviceStruct(), InitKeyboardDeviceStruct(), or a similar 
routine (see below)
and "opening" the device if necessary.
If you return a non-zero (i.e., != Success) value from the DEVICE_INIT
call, that device will be considered unavailable. If either the main keyboard
or main pointing device cannot be initialized, the DIX code will refuse 
to continue booting up.
.IP \(bu 5
DEVICE_ON - If the DeviceProc is called with DEVICE_ON, then it is 
allowed to start
putting events into the client stream by calling through the ProcessInputProc
in the device.
.IP \(bu 5
DEVICE_OFF - If the DeviceProc is called with DEVICE_OFF, no further 
events from that
device should be given to the DIX layer.
The device will appear to be dead to the user.
.IP \(bu 5
DEVICE_CLOSE - At DEVICE_CLOSE (terminate or reset) time, the device should
be totally closed down.
.LP
The sample server implementations for keyboard and mouse are in
server/ddx/dec/qvss/qvss_io.c.
.nf

	void InitPointerDeviceStruct(device, map, mapLength, GetMotionEvents, ControlProc)
		DevicePtr device;
		BYTE *map;
		int mapLength;
		void (*ControlProc)();
		int (*GetMotionEvents)();
.fi
InitPointerDeviceStruct is a DIX routine you call at DEVICE_INIT time to declare
some operating routines and data structures for a pointing device.
map and mapLength are as described in the X Window 
System protocol specification.
ControlProc and GetMotionEvents are DDX routines, see above.
.nf

	void InitKeyboardDeviceStruct(device, pKeySyms, pModifiers, Bell, ControlProc)
		DevicePtr device;
		KeySymsPtr pKeySyms;
		CARD8 *pModifiers;   
		void (*Bell)();
		void (*ControlProc)();

.fi
You call this DIX routine when a keyboard device is initialized and 
its device procedure is called with
DEVICE_INIT.
The formats of the keysyms and modifier maps are defined in 
server/include/input.h. 
They describe the layout of keys on the keyboards, and the glyphs 
associated with them.  ( See the next section for information on
setting up the modifier map and the keysym map.)
ControlProc and Bell are DDX routines, see above.

.NH 3
Keyboard Mapping and Keycodes
.XS
Keyboard Mapping and Keycodes
.XE
.LP
When you send a keyboard event, you send a report that a given key has either
been pressed or has been released.
There must be a keycode for each key that identifies the key;
the keycode-to-key mapping can be any mapping you desire, because you
specify the mapping in a table you set up for DIX.

The keycode mapping information that you set up consists of the following:
.IP \(bu 5
A minimum and maximum keycode number
.IP \(bu 5
An array of sets of glyphs for each key, that is of length 
maxkeycode - minkeycode + 1.  
Each element of this array is a list of codes for glyphs that are on that key.
There is no limit to the number of glyphs that can be on a key.
.LP
The sample server sets up the keycode and modifier maps in 
server/ddx/dec/lk201/lk201.c.  This is the map passed to the routine
InitKeyboardDeviceStruct().   Once the map is set up, DIX keeps and
maintains the client's changes to it.  See also server/ddx/dec/lk201/keynames.h
for the association between lk201 keys and the keysyms in X11/keysym.h.

The X protocol defines standard keycap glyph markings to indicate the 
symbol(s) printed on
each keycap. (See X11/keysym.h)

For instance, one of the Apple Macintosh keyboards has keys numbered from 0 to 58.
Key 0 has on it the glyph "A."  Key 1 has on it the glyph "S."
Key 41 has the glyphs ";" on the bottom and ":" on the top.
Key 48 is "Tab."
Key 49 is the space bar.
Key 56 is the shift keys (both keys return the same keycode).
An optional keypad generates codes 66 through 92, with some gaps in the middle.

Normally, the Macintosh system software translates these into ASCII for the 
application program.
An X server implementation would ignore the ASCII and just use the 
raw key codes.

Each glyph code is two bytes.
Given a space of 64K glyphs, the designers have used as much of this space 
as possible.
Whatever glyph is on your keyboard, there should be an appropriate glyph code
for it.
Fourteen glyph codes are for modifiers, including shift, control, Meta, Alt,
Super, and Hyper in both left and right flavors.
There are glyphs codes for the usual Return, Backspace, Rubout, Tab, etc.
There are codes for the 0 through 9 on the keypad as distinct from on the the 
regular keyboard,
besides the other glyphs commonly found on keypads.
There are cursor arrows and other control glyphs, such as Page Up, End, Home, 
Select, 
Undo, Help.
There are codes for PF1 through PF4.
They are distinct from F1 through F20.
There are, of course, the glyphs for the capital letters A through Z and all of the
punctuation marks that you have ever seen on any keyboard, 
including the division sign, cents sign, copyright, yen, and angle quotes.
In addition, glyph codes exist for all lowercase letters
and a huge
selection of letters with diacritical marks, ranging 
from a pretty typical 
uppercase N with a ~ 
over it to the lowercase d with a caron (upside-down circumflex).
There are diphthongs like ae and oe,
the German sharp S, and the Icelandic letter "eth," 
all in uppercase and lowercase.
These are all defined in X11/keysym.h.

Legal modifier keys must generate both up and down transitions.  When 
a client tries to change a modifier key (for instance, to make "A" the
"Control" key), DIX calls the following routine (in server/ddx/dec/lk201/lk201.c)
.nf

	Bool LegalModifier(key)
	    BYTE key;
.fi


.NH 2
Screens
.XS
Screens
.XE
.LP
Different computer graphics
displays have different capabilities.  
Some are simple monochrome
frame buffers that are just lying
there in memory, waiting to be written into.
Others are color displays with many bits per pixel using some color lookup table.
Still others have high-speed graphic processors that prefer to do all of the work 
themselves,
including maintaining their own high-level, graphic data structures.

.NH 3
Screen Hardware Requirements
.XS
Screen Hardware Requirements
.XE
.LP
The only requirement on screens is that you be able to both read
and write locations in the frame buffer.
All screens must have a depth of 32 or less (unless you use
an X extension to allow a greater depth).
All screens must fit into one of the classes listed in the section 
in this document on Visuals and Depths.
.LP
X uses the pixel as its fundamental unit of distance on the screen.
Therefore, most programs will measure everything in pixels.  
.LP
The sample server assumes square pixels.  
Serious WYSIWYG (what you see is what you get) applications for
publishing and drawing programs will adjust for
different screen resolutions automatically.
Considerable work
is involved in compensating for non-square pixels (both in the mfb
code for the sample server and the client applications).

.NH 3
Data Structures
.XS
Data Structures
.XE
.LP
X supports multiple screens that are connected to the same
server.  Therefore, all the per-screen information is bundled into one data
structure of attributes and procedures, which is the ScreenRec (see 
server/include/scrnintstr.h).  
The procedure entry points in a ScreenRec operate on 
regions, colormaps, cursors, and fonts, because these resources
can differ in format from one screen to another.

"Windows" are rectangular graphic areas on the screen 
that can be drawn into by graphic routines.
"Pixmaps" are off-screen graphic areas that can be drawn into.
They are both considered drawables and are 
described in the section on Drawables.
All graphic operations work on drawables,
and operations are available to copy patches from
one drawable to another.

The pixel image data in all drawables is in a format that is private
to DDX.
In fact, each instance of a drawable is associated with a given screen.
Presumably, the pixel image data for
pixmaps is chosen to be conveniently understood by the
hardware.   All screens in a single server must be able to handle 
all pixmaps depths declared in the connection setup information.
.LP
Pixmap images are transferred to the server in one of two ways:
XYPixmap or ZPimap.
XYPixmaps are a series of bitmaps, one for each bit plane of the image,
using the bitmap padding rules from the connection setup.
ZPixmaps are a series of nibbles, bytes or words, one for each pixel, 
using the format rules (padding and so on) for the appropriate depth.
.LP
All screens in a given server must agree on a set of
pixmap image formats (PixmapFormat) to support (depth, 
number of bits per pixel, etc.).
.LP
There is no color interpretation of bits in the pixmap.  Pixmaps 
do not contain pixel values.  The interpretation is made only when
the bits are transferred onto the screen.
.LP
The screenInfo structure (in scrnintstr.h) is a global data structure that
has a pointer to an array of ScreenRecs, one for each screen on the server.
(These constitute the one and only description of each screen in the server.)
Each screen has an identifying index (0, 1, 2, ...).
In addition, the screenInfo struct contains global server-wide
details, such as the bit- and byte-
order in all bit images, and the list of pixmap image formats that are supported.
Obviously, these must be the same for all screens on the server.

.NH 3
Output Initialization
.XS
Output Initialization
.XE
.LP
.nf

	InitOutput(pScreenInfo, argc, argv)
		ScreenInfo *pScreenInfo;
		int argc;
		char **argv;
.fi
Upon initialization, your DDX
routine InitOutput() is called by DIX.
It is passed a pointer to screenInfo to initialize.
It is also passed the argc and argv from main() for your server
for the command-line arguments.
These arguments may indicate what or how many screen device(s) to use
or in what way to use them.
For instance, your server command line may allow a "-D" flag 
followed by the name of the screen device to use.

Your InitOutput() routine should initialize each screen you wish to use
by calling AddScreen(), and then it should initialize the pixmap formats
that you support by storing values directly into the screenInfo data structure.
You should also set certain implementation-dependent numbers and 
procedures in your
screenInfo, which determines
the pixmap and scanline padding rules for all screens in the server.
(See the QVSS sample server implementation in server/ddx/ddx/qvss/init.c.)
.nf

	int AddScreen(scrInitProc, argc, argv)
		int (*scrInitProc)();
		int argc;
		char **argv;
.fi
You should call AddScreen(), a DIX procedure, in InitOutput()
once for each screen to add it to the 
screenInfo database.
The first argument is an initialization procedure for the screen that you supply.
The second and third are the argc and argv from main().
It returns the number of screens installed so far,
or -1 if there is insufficient memory to add the screen.

The scrInitProc should be of the following form:
.nf

	int scrInitProc(iScreen, pScreen, argc, argv)
		int iScreen;
		ScreenPtr pScreen;
		int argc;
		char **argv;
.fi
iScreen is the index for this screen; 0 for the first one initialized, 
1 for the second, etc.
pScreen is the pointer to the screen's new ScreenRec.
argc and argv are as before.
Your screen initialize procedure should return non-zero upon success or
zero if the screen
cannot be initialized (for instance, if the screen hardware does not exist on
this machine).

This procedure must determine what actual device it is supposed to initialize.
If you have a different procedure for each screen, then it is no problem.
If you have the same procedure for multiple screens, it may have trouble
figuring out which screen to initialize each time around, especially if
InitOutput() does not initialize all of the screens.
It is probably easiest to have one procedure for each screen.

The initialization procedure should fill in all the screen procedures
for that screen (windowing functions, region functions, etc.) and certain
screen attributes for that screen.
.LP
See server/ddx/dec/qvss/init.c and server/ddx/mfb/mfbscrinit.c for more details.

.NH 3
Region Routines in the ScreenRec
.XS
Region Routines in the ScreenRec
.XE
.LP
A region is a dynamically allocated data structure that describes
an irregularly shaped piece of real estate in XY pixel space.
You can think of it as a set of pixels on the screen to be operated upon with
set operations such as AND and OR.
.LP
A region is frequently implemented as a list of rectangles or bitmaps that
enclose the
selected pixels.
Region operators control the "clipping policy," or the operations that 
work on regions.
(The sample server
uses YX-banded rectangles.
Unless you have something already implemented for your
graphics system, you should keep that implementation.)
The procedure pointers to the region operators are located in the
ScreenRec data structure.
The definition of a region can be found in the file 
server/include/regionstr.h.
The region code is found in server/ddx/mi/miregion.c.
DDX implementations using other region formats will
need to supply different versions of the region operators.

Since the list of rectangles is unbounded in size, part of the region data
structure is usually a large, dynamically allocated chunk of memory.
As your region operators calculate logical combinations of 
regions, these blocks may need to be reallocated by your region 
software.
For instance, in the sample server, a RegionRec has some header information
and a pointer to a dynamically allocated rectangle list.
Periodically, the rectangle list needs to be expanded with Xrealloc(),
whereupon the new pointer is remembered in the RegionRec.
.nf

	RegionPtr pScreen->RegionCreate( rect, size)
		BoxPtr rect;
		int size;
.fi
RegionCreate creates a region that describes ONE rectangle.
The caller can avoid unnecessary reallocation and copying by declaring
the probable maximum number of rectangles that this region will need to 
describe itself.
Your region routines, though, cannot fail just because the region grows 
beyond this size.
The caller of this routine can pass almost anything as the size;
the value is merely a good guess as to the maximum size until it is proven
wrong by subsequent use.
Your region procedures are then on their own in
estimating how big the region will get.
Your implementation might ignore size, if applicable.
.nf

	void  pScreen->RegionCopy(dstrgn, srcrgn)
		RegionPtr dstrgn, srcrgn;
.fi
RegionCopy copies the description of one region, srcrgn, to another 
already-created region,
dstrgn.
.nf

	void pScreen->RegionDestroy( pRegion)
		RegionPtr pRegion;
.fi
RegionDestroy destroys a region and frees all allocated memory.
.nf

	int pScreen->Intersect(newReg, reg1, reg2)
		RegionPtr newReg, reg1, reg2;

	int  pScreen->Union(newReg, reg1, reg2)
		RegionPtr newReg, reg1, reg2;

	int  pScreen->Subtract(newReg, regMinuend, regSubtrahend)
		RegionPtr newReg, regMinuend, regSubtrahend;

	int pScreen->Inverse(newReg, pReg,  pBox)
		RegionPtr newReg, pReg;
		BoxPtr pBox;
.fi
The above four calls all do basic logical operations on regions.
They set the new region (which already exists)
to describe the logical intersection, union, set difference,
or inverse of the region(s) that were passed in.
Your routines must be able to handle a situation where the newReg is 
the same region as one of the other region arguments.

The subtract function removes the Subtrahend from the Minuend and
puts the result in newReg.

The inverse function returns a region that is the pBox minus the region passed in.
(A true "inverse" would make a region that extends to infinity in all directions
but has holes in the middle.)
It is undefined for situations where the region extends beyond the box.

Each routine must return the value TRUE for success.
.nf

	void pScreen->RegionReset(pRegion, pBox)
		RegionPtr pRegion;
		BoxPtr pBox;
.fi
RegionReset sets the region to describe
one rectangle and reallocates it to a size of one rectangle, if applicable.
.nf

	void  pScreen->TranslateRegion(pRegion, x, y)
		RegionPtr pRegion;
		int x, y;
.fi
TranslateRegion simply moves a region +x in the x direction and +y in the y 
direction.
.nf

	int  pScreen->RectIn(pRegion, pBox)
		RegionPtr pRegion;
		BoxPtr pBox;
.fi
RectIn returns one of the defined constants
rgnIN, rgnOUT, or rgnPART, depending upon whether the box is entirely
inside the region, entirely outside of the region, or partly in and partly out of 
the region.
These constants are defined in server/include/region.h.
.nf

	Bool pScreen->PointInRegion(pRegion, x, y, pBox)
		RegionPtr pRegion;
		int x, y;
		BoxPtr pBox;
.fi
PointInRegion returns true if the point x, y is in the region.
In addition, it fills the rectangle pBox with coordinates of a rectangle
that is entirely inside of pRegion and encloses the point.
In the mi implementation, it is the largest such rectangle.
(Due to the sample server implementation,
this comes cheaply.)

This routine used by DIX when tracking the pointing device and deciding whether
to report mouse events or change the cursor.
For instance, DIX needs to change the cursor when it moves from one window to
another.  Due to overlapping windows, the shape to check may be irregular.
A PointInRegion() call for every pointing device movement may be too expensive.
The pBox is a kind of wake-up box;
DIX need not call PointInRegion() again until the cursor wanders outside of 
the returned box.
.nf

	Bool pScreen->RegionNotEmpty(pRegion)
		RegionPtr pRegion;
.fi
RegionNotEmpty is a boolean function that returns
true or false depending upon whether the region encloses any pixels.
.nf

	void pScreen->RegionEmpty(pRegion)
		RegionPtr pRegion;
.fi
RegionEmpty sets the region to be empty.
.nf

	BoxPtr pScreen->RegionExtents(pRegion)
		RegionPtr pRegion;
.fi
RegionExtents returns a rectangle that is the smallest
possible superset of the entire region.
The caller will not modify this rectangle, so it can be the one
in your region struct.
.nf
	void pScreen->SendGraphicsExpose(client,pRegion,drawable,major,minor)
		ClientPtr client;
		RegionPtr pRegion;
		XID drawable;
		unsigned char major;
		unsigned short minor;
.fi
SendGraphicsExpose dispatches a list of GraphicsExposure events which
span the region to the specified client.  If the region is empty, or
a NULL pointer, a NoExpose event is sent instead.
.NH 3
Cursor Routines for a Screen
.XS
Cursor Routines for a Screen
.XE
.LP
A cursor is the visual form tied to the pointing device.
The default cursor is an "X" shape, but the cursor can have any shape.
When a client creates a window, it declares what shape the cursor will be when it
strays into that window on the screen.

For each possible shape the cursor assumes, there is a CursorRec data structure.
This data structure contains a bitmap for the image of the cursor and a 
bitmap for a mask
behind the cursor, in addition to foreground and background colors for the 
cursor.
The cursor image is applied to the screen by applying the mask first, 
clearing 1 bits in its
form to the background color, and then overwriting on the source image, in
the foreground color.
(One bits of the source image that fall on top of zero bits of the mask image
are undefined.)
This way, a cursor can have transparent parts, and opaque parts
in two colors.
X allows any cursor size, but some hardware cursor schemes allow a maximum
of N pixels by M pixels.
Therefore, you are allowed to transform the cursor to a smaller size, but be
sure to include the hot-spot.

CursorRec in server/include/cursorstr.h
is a device-independent structure containing a device-independent
representation of the bits for the source and mask.  
(This is possible because the bitmap representation is the same
for all screens.)

When a cursor is created, it is "realized" for
each screen.  At realization time, each screen has the chance to convert
the bits into some other representation that may be more convenient 
(for instance, putting the cursor
into off-screen memory) and
set up its device-private area in the cursor data structure to possibly point to 
whatever data
structures are needed.
For instance, the following
is the
device private entry for a particular screen and cursor:
.nf

	pCursor->devPrivate[pScreen->myNum]

.fi
This is done because the change from one cursor shape to another must
be fast and responsive;
the cursor image should be able to flutter as fast as the user moves it 
across the screen.

The sample server implementations of most of these routines are in 
server/ddx/dec/qvss/qvss_io.c.

You must implement the following routines for your hardware:
.nf

	Bool pScreen->RealizeCursor( pScr, pCurs)
		ScreenPtr pScr;
		CursorPtr pCurs;

	Bool pScreen->UnrealizeCursor( pScr, pCurs)
		ScreenPtr pScr;
		CursorPtr pCurs;
.fi
RealizeCursor and UnrealizeCursor
should realize (allocate and calculate all data needed) 
and unrealize (free the dynamically allocated data)
a given cursor when DIX needs them.
They are called whenever a device-independent
cursor is created or destroyed.
The source and mask bits pointed to by fields in pCurs are
undefined for bits beyond the right edge
of the cursor.  This is so because the bits are in Bitmap format, 
which may have pad bits on the right edge.
You should inhibit UnrealizeCursor() if the cursor is currently in use;
this happens when the system is reset.
.nf

	Bool pScreen->DisplayCursor( pScr, pCurs)
		ScreenPtr pScr;
		CursorPtr pCurs;
.fi
DisplayCursor should change the cursor on the given screen to the one passed in.
It is called by DIX when the user moves the pointing device into a 
different window with
a different cursor.  The hotspot in the cursor should be aligned
with the current cursor position.
.nf

	void pScreen->RecolorCursor( pScr, pCurs, displayed)
		ScreenPtr pScr;
		CursorPtr pCurs;
		int displayed;
.fi
.LP
RecolorCursor notifies DDX that the colors in pCurs have changed and
indicates whether this is the cursor currently being displayed.  If it
is, the cursor hardware state may have to be updated.  Whether
displayed or not, state created at RealizeCursor time may have to be
updated.  A generic version, miRecolorCursor, may be used that 
does an unrealize, a realize, and possibly a display (in micursor.c).
.nf

	void pScreen->ConstrainCursor( pScr, pBox)
		ScreenPtr pScr;
		BoxPtr pBox;
.fi
ConstrainCursor should cause the cursor to restrict its motion to the 
rectangle pBox.
DIX code is capable of enforcing
this constraint by forcefully moving the cursor if it strays out of the rectangle,
but ConstrainCursor offers a way to send a
hint to the driver or hardware if such support is available.  This can prevent the
cursor from wandering out of the box, then jumping back, as DIX forces it back.
.nf

	void pScreen->PointerNonInterestBox( pScr, pBox)
		ScreenPtr pScr;
		BoxPtr pBox;
.fi
PointerNonInterestBox is DIX's way of telling the pointing device code
not to report motion events while the cursor is inside a 
given rectangle on the given screen.
It is optional and, if not implemented, it should do nothing.
This routine is called only when the client has declared that it is 
not interested in motion events
in a given window.
The rectangle you get may be a subset of that window.
It saves DIX code the time required to discard uninteresting
mouse motion events.  This is only a hint, which may speed
performance.
.nf

	void pScreen->CursorLimits( pScr, pCurs, pHotBox, pTopLeftBox)
		ScreenPtr pScr;
		CursorPtr pCurs;
		BoxPtr pHotBox;
		BoxPtr pTopLeftBox;	/* return value */
.fi
.LP
CursorLimits should calculate the box that the cursor 
hot spot is
physically capable of moving within, as a function of the screen pScr,
the device-independent cursor pCurs, and a box that 
DIX hypothetically would want 
the hot spot
confined within, pHotBox.  
This routine is for informing DIX only; it alters no state within
DDX.
.nf

	Bool pScreen->SetCursorPosition( pScr, newx, newy, generateEvent)
		ScreenPtr pScr;
		unsigned int newx;
		unsigned int newy;
		Bool generateEvent;
.fi
.LP
SetCursorPosition should artificially move the cursor as though the user
had jerked the pointing device very quickly.
This is called in response to the WarpPointer request from the client,
and at other times.
If generateEvent is True, the device should decide whether or
not to call ProcessInputEvents() nd then it must call
DevicePtr->processInputProc.
Its effects are, of course, limited in value for absolute pointing devices such 
as a tablet.

.NH 3
Visuals, Depths and Pixmap Formats for Screens
.XS
Visuals, Depths and Pixmap Formats for Screens
.XE
.LP
The "depth" of a image is the number of bits that are used per pixel to display it.

The "bits per pixel" of a pixmap image that is sent over the client byte stream
is a number that is either 4, 8, 16, 24 or 32.
It is the number of bits used per pixel in Z format.
For instance, a pixmap image that has a depth of six is best sent
in Z format as 8 bits per pixel.

A "pixmap image format" or a "pixmap format"
is a description of the format of a pixmap image as it 
is sent over
the byte stream.
For each depth available on a server, there is one and only one 
pixmap format.
This pixmap image format gives the bits per pixel and the scanline padding
unit. (For instance, are pixel rows padded to 
bytes, 16-bit words, or 32-bit words?)

For each screen, you must decide upon what depth(s) it supports.
You should only count the number of bits used for the actual image.
Some displays store additional bits to indicate what window
this pixel is in, how close this object is to a viewer, transparency, 
and other data; do not count these bits.

A "display class" tells whether
the display is monochrome or color, whether 
there is a lookup table, and how the lookup table
works.

A "visual" is a combination of depth, display class,
and a description of how the pixel values result in a color on the screen.
Each visual has a set of masks and offsets that are used to separate a 
pixel value into its
red, green, and blue components and a count of the number of colormap entries.
Some of these fields are only meaningful when the class dictates so.
Each visual also has a screen ID telling which screen it is usable on.
Note that the depth does not imply the number of map_entries;
for instance, a display can have 8 bits per pixel but only 254 colormap entries
for use by applications (the other two being reserved by hardware for the cursor).

Each visual is identified by a 32-bit visual ID which the client uses to
choose what visual is desired on a given window.
Clients can be using more than one visual on the same screen at the same time;
.LP
The class of a display describes how this translation takes place.
There are three ways to do the translation.
.IP \(bu 5
Pseudo - The pixel value, as a whole, is looked up 
in a table of length map_entries to
determine the color to display.
.IP \(bu 5
True - The 
pixel value is broken up into red, green, and blue fields, each of which 
are looked up in separate red, green, and blue lookup tables, 
each of length map_entries.
.IP \(bu 5
Gray - The pixel value is looked up in a table of length map_entries to 
determine a gray level to display.
.LP
In addition, the lookup table can be static (resulting colors are fixed for each 
pixel value)
or dynamic (lookup entries are under control of the client program).
This leads to a total of six classes:

.IP \(bu 5
Static Gray - The pixel value (of however many bits) determines directly the 
level of gray
that the pixel assumes.  
.IP \(bu 5
Gray Scale - The pixel value is fed through a lookup table to arrive at the level 
of gray to display
for the given pixel.  
.IP \(bu 5
Static Color - The pixel value is fed through a fixed lookup table that yields the 
color to display
for that pixel.
.IP \(bu 5
PseudoColor - The whole pixel value is fed through a programmable lookup 
table that has one
color (including red, green, and blue intensities) for each possible pixel value,
and that color is displayed.
.IP \(bu 5
True Color - Each pixel value consists of one or more bits
that directly determine each primary color intensity after being fed through 
a fixed table.
.IP \(bu 5
Direct Color - Each pixel value consists of one or more bits for each primary color.
Each primary color value is individually looked up in a table for that primary 
color, yielding
an intensity for that primary color.
For each pixel, the red value is looked up in the
red table, the green value in the green table, and
the blue value in the blue table.
.LP
Here are some examples:
.IP
A simple monochrome 1 bit per pixel display is Static Gray.

A display that has 2 bits per pixel for a choice
between the colors of black, white, green and violet is Static Color.

A display that has three bits per pixel, where 
each bit turns on or off one of the red, green or
blue guns, is in the True Color class.

If you take the last example and scramble the
correspondence between pixel values and colors
it becomes a Static Color display.

A display has 8 bits per pixel.  The 8 bits select one entry out of 256 entries
in a lookup table, each entry consisting of 24 bits (8bits each for red, green,
and blue).
The display can show any 256 of 16 million colors on the screen at once.
This is a pseudocolor display.
The client application gets to fill the lookup table in this class of display.

Imagine the same hardware from the last example.
Your server software allows the user, on the 
command line that starts up the server
program, 
to fill the lookup table to his liking once and for all.
From then on, the server software would not change the lookup table
until it exits.
For instance, the default might be a lookup table with a reasonable sample of 
colors from throughout the color space.
But the user could specify that the table be filled with 256 steps of gray scale
because he knew ahead of time he would be manipulating a lot of black-and-white 
scanned photographs
and not very many color things.
Clients would be presented with this unchangeable lookup table.
Although the hardware qualifies as a PseudoColor display,
the facade presented to the X client is that this is a Static Color display.

You have to decide what kind of display you have or want
to pretend you have.  
When you initialize the screen(s), this class value must be set in the
VisualRec data structure along with other display characteristics like the 
depth and other numbers.

The allowable DepthRec's and VisualRec's are pointed to by fields in the ScreenRec.
These are set up when InitOutput() is called; you should Xalloc() appropriate blocks
or use static variables initialized to the correct values.

.NH 3
Colormaps for Screens
.XS
Colormaps for Screens
.XE
.LP
A colormap is a device-independent
mapping between pixel values and colors displayed on the screen.

Different windows on the same screen can have different
colormaps at the same time.
At any given time, the most recently installed
colormap(s) will be in use in the server
so that its (their) windows' colors will be guaranteed to be correct.
Other windows may be off-color.
Although this may seem to be chaotic, in practice most clients 
use the default colormap for the screen.

The default colormap for a screen is initialized when the screen is initialized.
It always remains in existence and is not owned by any regular client.  It 
is owned by client 0 (the server itself).
Many clients will simply use this default colormap for their drawing.
Depending upon the class of the screen, the entries in this colormap may
be modifiable by client applications.

.NH 4
Colormap Routines
.XS
Colormap Routines
.XE
.LP
You need to implement the following routines to
handle the device-dependent aspects of color maps.
You will end up placing pointers to these procedures
in your ScreenRec data structure(s).
The sample server implementations of many of these routines are in mfbcmap.c;
since mfb does not do very much with color, many of these routines are
set to no-op procedures.
.nf

	Bool pScreen->CreateColormap(pColormap)
		ColormapPtr pColormap;
.fi
.LP
This routine is called by the DIX CreateColormap routine after it has allocated
all the data for the new colormap and just before it returns to the dispatcher.
It is the DDX layer's chance to initialize the colormap, particularly if it is
a static map.  See the following
section for more details on initializing colormaps.
The routine returns FALSE if creation failed, such as due to memory
limitations.
Notice that the colormap has a devPriv field from which you can hang any
colormap specific storage you need.  Since each colormap might need special
information, we attached the field to the colormap and not the visual.
.nf

	pScreen->DestroyColormap(pColormap)
		ColormapPtr pColormap;
.fi
.LP
This routine is called by the DIX FreeColormap routine after it has uninstalled
the colormap and notified all interested parties, and before it has freed
any of the colormap storage.
It is the DDX layer's chance to free any data it added to the colormap.
.nf

	pScreen->InstallColormap(pColormap)
		ColormapPtr pColormap;
.fi
.LP
InstallColormap should 
fill a lookup table on the screen with which the colormap is associated with
the colors in pColormap.
If there is only one hardware lookup table for the screen, then all colors on
the screen may change simultaneously.

In the more general case of multiple hardware lookup tables,
this may cause some other colormap to be
uninstalled, meaning that windows that subscribed to the colormap
that was uninstalled may end up being off-color.
See the note, below, about uninstalling maps.
.nf

	pScreen->UninstallColormap(pColormap)
		ColormapPtr pColormap;
.fi
.LP
UninstallColormap should 
remove pColormap from screen pColormap->pScreen.  
Some other map, such as the default map if possible,
should be installed in place of pColormap if applicable.
If
pColormap is the default map, do nothing.
If any client has requested ColormapNotify events, the DDX layer must notify the client.  
(The routine WalkTree() is 
be used to find such windows.  The DIX routines TellNoMap(), 
TellNewMap()  and TellGainedMap() are provided to be used as 
the procedure parameter to WalkTree.  These procedures are in
server/dix/colormap.c.)
.nf

	int pScreen->ListInstalledColormaps(pScreen, pCmapList)
		ScreenPtr pScreen;
		Colormap *pCmapList;
.fi
.LP
ListInstalledColormaps fills the pCMapList in with the resource ids
of the installed maps and returns a count of installed maps.
pCmapList will point to an array of size MaxInstalledMaps that was allocated
by the caller.
.nf

	void pScreen->StoreColors (pmap, ndef, pdefs)
		ColormapPtr pmap;
		int ndef;
		xColorItem *pdefs;
.fi
.LP
StoreColors changes some of the entries in the colormap pmap.
The number of entries to change are ndef, and pdefs points to the information
describing what to change.
Note that partial changes of entries in the colormap are allowed.
Only the colors
indicated in the flags field of each xColorItem need to be changed.  
However, all three color fields will be sent with the proper value for the
benefit of screens that may not be able to set part of a colormap value.
If the screen is a static class, this routine does nothing.
The structure of colormap entries is nontrivial; see colormapst.h 
and the definition of xColorItem in Xproto.h for 
more details.
.nf

	void pScreen->ResolveColor(pRed, pGreen, pBlue, pVisual)
		unsigned short *pRed, *pGreen, *pBlue;
		VisualPtr pVisual;

.fi
.LP
Given a requested color, ResolveColor returns the nearest color that this hardware is
capable of displaying on this visual.
In other words, this rounds off each value, in place, to the number of bits
per primary color that your screen can use.
Remember that each screen has one of these routines.
The level of roundoff should be what you would expect from the value
you put in the bits_per_rgb field of the pVisual.

Each value is an unsigned value ranging from 0 to 65535.
The bits least likely to be used are the lowest ones.
.LP
For example, if you had a pseudocolor display
with any number of bits per pixel
that had a lookup table supplying 6 bits for each color gun
(a total of 256K different colors), you would
round off each value to 6 bits.
.NH 4
Initializing a Colormap
.XS
Initializing a Colormap
.XE
.LP
When a client requests a new colormap and when the server creates the default
colormap, the procedure CreateColormap in the DIX layer is invoked.
That procedure allocates memory for the colormap and related storage such as
the lists of which client owns which pixels.  
It then sets a bit, BeingCreated, in the flags field of the ColormapRec
and calls the DDX layer's CreateColormap routine.
This is your chance to initialize the colormap.
If the colormap is static, which you can tell by looking at the class field,
you will want to fill in each color cell to match the hardwares notion of the
color for that pixel.
If the colormap is the default for the screen, which you can tell by looking
at the IsDefault bit in the flags field, you should allocate BlackPixel
and WhitePixel to match the values you set in the pScreen structure.
(Of course, you picked those values to begin with.)
.LP
You can also wait and use AllocColor() to allocate blackPixel 
and whitePixel after the default colormap has been created.
If the default colormap is static and you initialized it in
pScreen->CreateColormap, then use can use AllocColor afterwards
to choose pixel values with the closest rgb values to those
desired for blackPixel and whitePixel.
If the default colormap is dynamic and uninitialized, then
the rgb values you request will be obeyed, and AllocColor will
again choose pixel values for you.
These pixel values can then be stored into the screen.
.LP
There are two ways to fill in the colormap.
The simplest way is to use AllocColor.  
.nf

AllocColor (pmap, pred, pgreen, pblue, pPix, client)
    ColormapPtr         pmap;
    unsigned short      *pred, *pgreen, *pblue;
    Pixel               *pPix;
    int                 client;

.fi
This takes three pointers to 16 bit color values and a pointer to a suggested
pixel value.  The pixel value is either an index into one colormap or a
combination of three indices depending on the type of pmap.
If your colormap starts out empty, and you don't deliberately pick the same
value twice, you will always get your suggested pixel.
The truly nervous could check that the value returned in *pPix is the one
AllocColor was called with.
If you don't care which pixel is used, or would like them sequentially
allocated from entry 0, set *pPix to 0.  This will find the first free
pixel and use that.
.LP
AllocColor will take care of all the  bookkeeping  and  will
call StoreColors to get the colormap rgb values initialized.
The hardware colormap will be changed whenever this colormap
is installed.
.LP
If for some reason AllocColor doesn't do what you want, you can do your
own bookkeeping and call StoreColors yourself.  This is much more difficult
and shouldn't be necessary for most devices.

.NH 3
Fonts for Screens
.XS
Fonts for Screens
.XE
.LP
A font is a set of bitmaps that depict the symbols in a character set.
Each font is for only one typeface in a given size, in other words, just one
bitmap for each character.
Parallel fonts may be available in a variety of sizes and variations, including
"bold" and "italic."
X supports fonts for 8-bit and 16-bit character codes (for oriental languages
that have more than 256 characters in the font).
Glyphs are bitmaps for individual characters.

The source comes with some useful font files in an
ASCII, plain-text format that should be comprehensible on a wide variety of operating systems.
The text format, referred to as BDF, is a slight extension of the
current Adobe 2.1 Bitmap Distribution Format (Adobe Systems, Inc.).

A short paper in PostScript format is included with the sample server
that defines BDF.  It includes helpful pictures, which is why it is
done in PostScript and is not included in this document.

Your implementation should include some sort of font compiler to read these
files and generate binary files that are directly usable by your server implementation.
The sample server comes with the source for a font compiler.

It is important the font properties contained in the BDF files are
preserved across any font compilation. In particular, copyright
information cannot be casually tossed aside without legal
ramifications. Other properties will be important to
some sophisticated applications.

All clients get font information from the server.
Therefore, your server can support any fonts it wants to.
It should probably support at least the fonts supplied with the X11 tape.
In principle, you can convert fonts from other
sources or dream up your own fonts for use on your server.

.NH 4
Server Natural Format
.XS
Server Natural Format
.XE
.LP
A font compiler is supplied with the sample server.
It has compile-time switches to convert the BDF files
into a simple binary form, called Server Natural Format or SNF,
with all the bit- and
byte-swapping issues being resolved.
The font compiler should be directly portable to most UNIX-based systems 
and is probably portable
to many non-UNIX systems.

WARNING: the bit and byte order defines
in the font compiler source are distinct from those for
the main server code.  This is for you to cross-compile
a font file for a different machine.
If these are set differently for the same server then your text will
not draw correctly.

The fonts included with the tape are stored in fonts/bdf.  The
font compiler is found in fonts/compiler.

Server Natural Format font files consist of five parts. All of the data
structures in the file are declared in server/include/font.h. Each part begins
on a 32-bit (4 byte) boundary in the file.   The structures in
server/include/font.h are shared by the font compiler and server.

The FontInfoRec part is a header that has global information common to all 
characters in the font and the sizes of the other parts of the file.
This structure contains the data necessary for the server to
respond to a QueryFontRequest. The structure contains a version number
(FONT_FILE_VERSION) at the beginning and the end of the structure to
aid in detecting bad or outdated files.

The Character Information part has metrics for each character in the font.
It is an array of character metrics. The first
element describes the character at fi.firstRow and fi.firstCol.
In
addition to the protocol defined metrics, the XCHARINFO
structure contains a ci.byteOffset and ci.bitOffset for each character.
The ci.byteOffset offset from the character glyph at the
beginning of the glyphs (below). The ci.bitOffset is the sum of the
bounding box widths of all preceding characters.

The Character Glyphs part has bit images for each character in the font.
Each scanline of each glyph is padded to a byte boundary
Bit and byte order is whatever is natural for the server.
(Note: the current BDF to SNF font compiler handles either bit order
within a byte as a compile time option. It does not need to
deal with byte order for the sample server implementation.)
The glyph for a character with XCHARINFO as ci begins at
cg[ci.byteOffset]. Glyphs may begin at arbitrary byte offsets within
the array.

The Property Descriptors part is a way to record arbitrary properties and metrics
that were not designed into the font scheme for X.
The pd.name field of each entry is the offset into the string table
of the null-terminated name of the property. If pd.indirect is FALSE,
pd.value is the INT32 value of the property. Otherwise, pd.value
is the offset of the null-terminated string property in the string
table.

The next part is the Property Strings (the values of the properties).
These are the null-terminated strings for property names and values.
For a property descriptor pd, the name is at s[pd.name] and the
value is at s[pd.value] if the pd.indirect is TRUE. All strings
are null-terminated and may begin at any byte offset from the
beginning.

The strings array is NOT padded to a 4-byte boundary.

.NH 4
Font Loading and Manipulation
.XS
Font Loading and Manipulation
.XE
.LP

There are a number of routines used by the server to deal with loading
and manipulating fonts.
.nf

	SetDefaultFont(defaultfontname)
		char *defaultfontname;
.fi
SetDefaultFont establishes the default font in graphics contexts.
.nf

	FontPtr FontFileLoad(pfontname, lenfname);
		char *pfontname;
		unsigned lenfname;
.fi
FontFileLoad is the primary routine for loading a font.
It should call ExpandFontName, and should try to share
any existing font structure for the font.  DIX is responsible for
calling RealizeFont for each screen.
.nf

	int FontUnload(pfont)
		FontPtr pfont;
.fi
FontUnload is used to unload a font.  It should free the
storage for the font.  DIX will have already called
UnrealizeFont for each screen.
.nf

	Bool FontFilePropLoad(fontname, length, font, fi, props)
		char *fontname;
		unsigned int length;
		FontInfoPtr fi;
		DIXFontPropPtr *props;
		FontPtr	*font;
.fi
FontFilePropLoad gets used by ListFontsWithInfo to extract
metric and property information from the font.  It should
return False if the font does not exist.  The font name
should be expanded with ExpandFontName.

Routines to deal with SNF format are
supplied (in server/ddx/snf), but you can replace or augment these if you want
to do conversion from other formats.  For each format, the OS layer needs to
know the file extension for the format (e.g. "snf"), an optional filter program
(given as a command line) to execute to convert the format to a usable format
(e.g., to uncompress a compressed file, or to compile a BDF file on the fly),
plus routines to load the font, load the font properties, and free the font.
.nf

	FontPtr FontFileReader->loadFont(fp)
		FID fp;
.fi
This routine takes an open file and loads the font.
.nf

	Bool FontFileReader->loadProperties(fp)
		FID fp;
		FontInfoPtr pfi;
		DIXFontPropPtr *ppdfp;
.fi
This routine takes an open file and loads the font properties and metrics.
.nf

	FontFileReader->freeFont(font)
		FontPtr font;
.fi
This routine frees the storage associated with a font.

.NH 4
Font Realization
.XS
Font Realization
.XE
.LP
Each screen configured into the server
has an opportunity at font-load time
to "realize" a font into some internal format if necessary. 
This happens every time the font is loaded into memory.

A font (FontRec in server/include/dixfontstr.h) is
a device-independent structure containing a device-independent
representation of the font.  When a font is created, it is "realized"
for each screen.  At this point, the screen has the chance to convert
the font into some other format.  The DDX layer can also put information
in the devPrivate storage.
.nf

	Bool pScreen->RealizeFont(pScr, pFont)
		ScreenPtr pScr;
		FontPtr pFont;

	Bool pScreen->UnrealizeFont(pScr, pFont)
		ScreenPtr pScr;
		FontPtr pFont;
.fi
RealizeFont and UnrealizeFont should calculate and allocate these extra data structures and 
dispose of them when no longer needed.
These are called in response to OpenFont and CloseFont requests from 
the client.
The sample server implementation is in mfbfont.c.

.NH 3
Other Screen Routines
.XS
Other Screen Routines
.XE
.LP
You must supply several other screen-specific routines for 
your X server implementation.
Some of these are described in other sections:
.IP \(bu 5
GetImage() is described in the Drawing Primitives section.
.IP \(bu 5
GetSpan() is described in the Pixblit routine section.
.IP \(bu 5
Several window and pixmap manipulation procedures are 
described in the Window section under Drawables.
.IP \(bu 5
The CreateGC() routine is described under Graphics Contexts.
.LP
.nf

	void pScreen->QueryBestSize(kind, pWidth, pHeight)
		int kind;
		CARD16 *pWidth, *pHeight;
.fi
QueryBestSize() returns the best sizes for cursors, tiles, and stipples
in response to client requests.
kind is one of the defined constants CursorShape, TileShape, or StippleShape
(defined in X.h).
For CursorShape, return the maximum width and 
height for cursors that you can handle.
For TileShape and StippleShape, start with the suggested values in pWidth
and pHeight and modify them in place to be optimal values that are
greater than or equal to the suggested values.
The sample server implementation is in server/ddx/mfb/mfbmisc.c.
.nf

	Bool pScreen->SaveScreen(pScreen, on)
		ScreenPtr pScreen;
		int on;
.fi
SaveScreen() is used for Screen Saver support (see WaitForSomething()).
pScreen is the screen to save.
See also server/ddx/dec/qvss/qvss_io.c, and server/os/4.2bsd/WaitFor.c.
.nf

	Bool pScreen->CloseScreen(pScreen)
	    ScreenPtr pScreen;
.fi
When the server is reset, it calls this routine for each screen.
.LP
As a convenience, the Screen structure contains an array of 
GCs that are preallocated, one at each depth the screen supports.
These are useful in the mi code.  Two routines must be used to
get these GC:
.nf

	GCPtr GetScratchGC(pScreen, depth)
	    ScreenPtr pScreen;
	    int depth;


	FreeScratchGC(pGC)
	    GCPtr pGC;
.fi
Always use these two routines, don't try to extract the scratch
GC yourself -- someone else might be using it, so a new one must
be created on the fly.
.LP
If you need a GC for a very long time, say until the server is restarted,
you should not take one from the pool used by GetScratchGC, but should
get your own from CreateScratchGC.  
This leaves the ones in the pool free for routines that only need it for
a little while and don't want to pay a heavy cost to get it.
.nf
	GCPtr CreateScratchGC(pScreen, depth)
	    ScreenPtr pScreen;
	    int depth;

.fi
NULL is returned if the GC cannot be created.
The GC returned can be freed with FreeScratchGC.
.NH 2
Drawables
.XS
Drawables
.XE
.LP
A drawable is a descriptor of a surface that graphics are drawn into, either
a window on the screen or a pixmap in memory.

Each drawable has a type,
ScreenPtr for the screen it is associated with, depth,
and serial number.
The type is one of the defined constants DRAWABLE_PIXMAP,
DRAWABLE_WINDOW and UNDRAWABLE_WINDOW.
(An undrawable window is used for window class InputOnly.)
The serial number is guaranteed to be unique across drawables, and
is used in determining
the validity of the clipping information in a GC.
The screen selects the set of procedures used to manipulate and draw into the
drawable.
There are, in fact, no other fields that a window drawable and pixmap
drawable have in common besides those mentioned here.

Both PixmapRecs and WindowRecs are  structs that start with a
drawable and continue on with more fields.
They have devPrivate pointers, which are assumed to point to everything else needed.
This is done because different graphics hardware has different requirements for
management;
if the graphics is always handled by a processor with an independent address space, there
is no point having a pointer to the bit image itself.

The definition of a drawable and a pixmap can be found in the file
server/include/pixmapstr.h.
The definition of a window can be found in the file server/include/windowstr.h.

.NH 3
Pixmaps
.XS
Pixmaps
.XE
.LP
A pixmap is a three-dimensional array of bits stored in memory, rather
than in the screen's display frame buffer.  It can
be used as a source or destination in graphics operations.
There is no implied interpretation of the pixel values in a pixmap, because it
has no associated visual or colormap.
There is only a depth that indicates the number
of significant bits per pixel.
Also, there is no implied physical size for each pixel; 
all graphic units are in numbers of pixels.
Therefore, a pixmap alone does not constitute a complete image;
it represents only a rectangular array of pixel values.

Note that the pixmap data structure is reference-counted.

The server
implementation is free to put the pixmap image data anywhere it sees fit,
according to its graphics hardware setup.
Many implementations will simply have the data dynamically
allocated in the server's address space.

The pixmap data structure has two fields that are private to the device.
Although you can use them for anything you want, they have intended purposes.
devKind is intended to be a device specific indication of the pixmap location (host
memory, off-screen, etc.).    In the
sample server, since all pixmaps are in memory, devKind stores the
width of the pixmap in bitmap scanline units.
devPrivate is probably a pointer to
the bits in the pixmap plus other device specific information.

A bitmap is a pixmap that is one bit deep.
.nf

	PixmapPtr pScreen->CreatePixmap(pScreen, width, height, depth)
		ScreenPtr pScreen;
		int width, height, depth;
.fi
This ScreenRec procedure must create a pixmap of the size
requested.
It must allocate a PixmapRec and fill in all of the fields.
The reference count field must be set to 1.
If successful, it returns a pointer to it; if not, it returns NULL.
See server/ddx/mfb/mfbpixmap.c for the server implementation.
.nf

	Bool pScreen->DestroyPixmap(pPixmap)
		PixmapPtr pPixmap;
.fi
This ScreenRec procedure must "destroy" a pixmap.
It should decrement the reference count and, if zero, it 
must deallocate the PixmapRec and all attached devPrivate blocks.
If successful, it returns TRUE. 
See server/ddx/mfb/mfbpixmap.c for the server implementation.

.NH 3
Windows
.XS
Windows
.XE
.LP
A window is a visible, or potentially visible, rectangle on the screen.
DIX windowing functions maintain an internal n-ary tree data structure, which
represents the current relationships of the mapped windows.
Windows that are contained in another window are children of that window and
are clipped to the boundaries of the parent.
The root window in the tree is the window for the entire screen.
Sibling windows constitute a doubly-linked list; the parent window has a pointer
to the head and tail of this list.
Each child also has a pointer to its parent.

The border of a window is drawn by a DDX procedure when DIX requests that it be drawn.
The contents of the window is drawn by the client through requests to the server.

Window painting is orchestrated through an expose event system.
When a region is exposed, 
DIX generates an expose event, telling the client to repaint the window and
passing the region that is the minimal area needed to be repainted.

As a favor to clients, the server may retain
the output to the hidden parts of windows
in off-screen memory; this is called "backing store".
When a part of such a window becomes exposed, it
can quickly move pixels into place instead of
triggering an expose event and waiting for a client on the other
end of the network to respond.
Even if the network response is insignificant, the time to
intelligently paint a section of a window is usually more than
the time to just copy already-painted sections.
At best, the repainting involves blanking out the area to a background color,
which will take about the
same amount of time.
In this way, backing store can dramatically increase the
performance of window moves.

On the other hand, backing store can be quite complex, because
all graphics drawn to hidden areas must be intercepted and redirected
to the off-screen window sections.
Not only can this be complicated for the server programmer,
but it can also impact window painting performance.
The backing store implementation can choose, at any time, to 
forget pieces of backing that are written into, relying instead upon
expose events to repaint for simplicity.

In X, the decision to use the backing-store scheme is made
by you, the server implementor.
X provides hooks for implementing backing store, therefore 
the decision to use this strategy can be made on the fly.
For example, you may use backing store only for certain windows
that the user requests or you may use backing store 
until memory runs out, at which time you
start dropping pieces of backing as needed to make more room.

When a window operation is requested by the client,
such as a window being created or moved,
a new state is computed.
During this transition, DIX informs DDX what rectangles in what windows are about to
become obscured and what rectangles in what windows have become exposed.
This provides a hook for the implementation of backing store.
If DDX is unable to restore exposed regions, DIX generates expose
events to the client.
It is then the client's responsibility to paint the
window parts that were exposed but not restored.

If a window is resized, pixels sometimes need to be
moved, depending upon
the application.
The client can request "Gravity" so that
certain blocks of the window are
moved as a result of a resize.
For instance, if the window has controls or other items
that always hang on the edge of the
window, and that edge is moved as a result of the resize,
then those pixels should be moved
to avoid having the client repaint it.
If the client needs to repaint it anyway, such an operation takes
time, so it is desirable
for the server to approximate the appearance of the window as best
it can while waiting for the client
to do it perfectly.
Gravity is used for that, also.

The window has several fields used in drawing
operations:
.IP \(bu 5
clipList - This region, in conjunction with
the client clip region in the gc, is used to clip output.
clipList has the window's children subtracted from it, in addition to pieces of sibling windows
that overlap this window.  To get the list with the
children included (subwindow-mode is IncludeInferiors),
the routine NotClippedByChildren(pWin) returns the unclipped region.
.IP \(bu 5
borderClip is the region used by CopyWindow and 
includes the area of the window, its children, and the border, but with the
overlapping areas of sibling children removed.
.IP \(bu 5
absCorner is the absolute screen coordinate
of the upper-left corner of this window.
.LP
Most of the other fields are for DIX use only.

.NH 4
Window Procedures in the ScreenRec
.XS
Window Procedures in the ScreenRec
.XE
.LP
You should implement
all of the following procedures and store pointers to them in the screen record.

The device-independent portion of the server "owns" the window tree.
However, clever hardware might want to know the relationship of
mapped windows.  There are pointers to procedures
in the ScreenRec data structure that are called to give the hardware
a chance to update its internal state.  These are helpers and
hints to DDX only;
they do not change the window tree, which is only changed by DIX.
.nf

	void pScreen->CreateWindow(pWin)
		WindowPtr pWin;
.fi
This routine is a hook for when DIX creates a window.
It should fill in the "Window Procedures in the WindowRec" below
and also allocate the devPrivate block for it.

See server/ddx/mfb/mfbwindow.c for the sample server implementation.
.nf

	Bool pScreen->DestroyWindow(pWin);
		WindowPtr pWin;
.fi
This routine is a hook for when DIX destroys a window.
It should deallocate the devPrivate block for it and any other blocks that need
to be freed, besides doing other cleanup actions.

See server/ddx/mfb/mfbwindow.c for the sample server implementation.
.nf

	Bool pScreen->PositionWindow(pWin, x, y);
		WindowPtr pWin;
		int x, y;
.fi
This routine is a hook for when DIX moves or resizes a window.
It should do whatever private operations need to be done when a window is moved or resized.
For instance, if DDX keeps a pixmap tile used for drawing the background
or border, and it keeps the tile rotated such that it is longword
aligned to longword locations in the frame buffer, then you should rotate your tiles here.
The actual graphics involved in moving the pixels on the screen and drawing the
border are handled by CopyWindow(), below.
.LP
See server/ddx/mfb/mfbwindow.c for the sample server implementation.
.nf

	void pScreen->RealizeWindow(pWin);
		WindowPtr pWin;

	void  pScreen->UnrealizeWindow(pWin);
		WindowPtr pWin;
.fi
These routines are hooks for when DIX maps (makes visible) and unmaps (makes invisible)
a window.
It should do whatever private operations need to be done 
when these happen, such as allocating or deallocating structures that 
are only needed for visible windows.
RealizeWindow does NOT draw the window border, background or contents;
UnrealizeWindow does NOT erase the window or generate exposure events
for underlying windows; this is taken care of by DIX.
DIX does, however, call PaintWindowBackground() and PaintWindowBorder()
to perform some of these.

.nf

	Bool pScreen->ChangeWindowAttributes(pWin, vmask)
		WindowPtr pWin;
		long vmask;
.fi

ChangeWindowAttributes is called whenever DIX changes window attributes, such as the
size, front-to-back ordering, title, or anything of lesser severity that
affects the window itself.
The sample server implements this routine.  It computes accelerators
for quickly putting up background and border tiles.  (See description of the
set of routines stored in the WindowRec.)
.nf

	int pScreen->ValidateTree(pParent,  pChild, top, anyMarked)
		WindowPtr pParent, pChild;
		Bool top, anyMarked;
.fi

ValidateTree calculates the clipping region for the parent window and
all of its children.
This routine must be provided. The sample
server has a machine-independent version in server/ddx/mi/mivaltree.c.
.nf

	void pScreen->WindowExposures(pWin, pRegion)
		WindowPtr pWin;
		RegionPtr pRegion;
.fi
The WindowExposures() routine
paints the border and generates exposure events for the window.
Since exposure events include a rectangle describing what was exposed, 
this routine may have to send back a series of exposure events, one for
each rectangle of the region.  
The count field in the expose event is a hint to the
client as to the number of
regions that are after this one.
This routine must be provided. The sample
server has a machine-independent version in server/ddx/mi/miexpose.c.

.NH 4
Window Procedures in the WindowRec
.XS
Window Procedures in the WindowRec
.XE
.LP
In addition to the procedures stored in the ScreenRec, several routines
are kept in the WindowRec itself.
In the sample server, mi implementations will work for 
most purposes and mfb routines speed up situations, such
as solid backgrounds/borders or tiles that are 8, 16 or 32 pixels square.

These three routines are used for systems that implement a backing-store scheme for it to
know when to stash away areas of pixels and to restore or reposition them.
.nf

	void pWindow->ClearToBackground(pWin, x, y, w, h, generateExposures);
		WindowPtr pWin;
		int x, y, w, h;
		Bool generateExposures;
.fi
This routine is called on a window in response to a ClearToBackground request
from the client.
This request has two different but related functions, depending upon generateExposures.

If generateExposures is true, the client is declaring that the given rectangle
on the window is incorrectly painted and needs to be repainted.
The sample server implementation calculates the exposure region
and hands it to the DIX procedure HandleExposures(), which
calls the WindowExposures() routine, below, for the window
and all of its child windows.

If generateExposures is false, the client is trying to simply erase part
of the window to the background fill style.
ClearToBackground should write the background color or tile to the 
rectangle in question (probably using PaintWindowBackground).
If w or h is zero, it clears all the way to the right or lower edge of the window.

The sample server implementation is in server/ddx/mi/miwindow.c.
.nf

	void pWindow->PaintWindowBackground(pWin, region, kind)
		WindowPtr pWin;
		RegionPtr region;
		int kind;	/* must be PW_BACKGROUND */

	void pWindow->PaintWindowBorder(pWin, region, kind)
		WindowPtr pWin;
		RegionPtr region;
		int kind;	/* must be PW_BORDER */
.fi
These two routines are for painting pieces of the window background or border.
They both actually paint the area designated by region.
The kind parameter is a defined constant that is always PW_BACKGROUND
or PW_BORDER, as shown.
Therefore, you can use the same routine for both.
The defined constant tells the routine whether to use the window's 
border fill style or its background fill style to paint the given region.
Both fill styles consist of a tile pointer and a pixel value. 
If the tile pointer is USE_PIXEL_VALUE, 
the background is the solid pixel value.
.nf

	void pWindow->CopyWindow(pWin, oldpt, oldRegion);
		WindowPtr pWin;
		POINT oldpt;
		RegionPtr oldRegion;
.fi

CopyWindow is called when a window is moved, and graphically
moves to pixels of a window on the screen.
It should not change any other state within DDX (see PositionWindow(), above).

oldpt is the old location of the upper-left corner.
oldRegion is the old region it is coming from.
The new location and new region is stored in the WindowRec.
oldRegion might modified in place by this routine (the sample
implementation does this).

CopyArea could be used, except that this operation has more complications.
First of all, you do not want to copy a rectangle onto a rectangle.
The original window may be obscured by other windows, and the new window location
may be similarly obscured.  
Second, some hardware supports multiple windows with multiple depths, and 
your routine needs to take care of that.

The pixels
in oldRegion (with reference point oldpt) are copied to the
window's new region (pWin->borderClip).   pWin->borderClip is gotten
directly from the window, rather than passing it as a parameter.

The sample server implementation is in server/ddx/mfb/mfbwindow.c.

.NH 4
Window Operations for Backing Store
.XS
Window Operations for Backing Store
.XE
.LP
Each WindowRec has a pointer to a struct of type BackingStoreRec.
For windows not supporting backing store, this pointer is null.
Servers that implement some backing store scheme must allocate 
a BackingStoreRec, must fill in the procedure pointers for the procedures below,
and must maintain a few fields in the WindowRec specifically for backing store.

These three routines are used for systems that 
implement a backing store scheme for it to
know when to stash away areas of pixels and to restore or reposition them.
.nf

	void pBackingStore->SaveDoomedAreas(pWin);
		WindowPtr pWin;
.fi

This routine looks at the obscured region of the window and tries to save
those pixels somewhere.
.nf

	RegionPtr pBackingStore->RestoreAreas(pWin);
		WindowPtr pWin;
.fi
This looks at the exposed region of the window.  
It tries to restore to the screen the parts that have been saved.
It removes the restored parts from the backing storage (because
they are now on the screen) and subtracts the areas from
the exposed region.  The returned region is the area of the window
which should have expose events generated for and can be either a new
region, pWin->exposed, or NULL.  The region left in pWin->exposed
is set to the area of the window which should be painted with
the backgroundTile.
.nf

	void pBackingStore->TranslateBackingStore(pWin, dx, dy);
		WindowPtr pWin;
		int dx, dy;
.fi
This is called when the window is moved or resized so that the backing
store can be translated if necessary.


.NH 2
Graphics Contexts and Validation
.XS
Graphics Contexts and Validation
.XE
.LP
This graphics context (GC) contains state variables such as foreground and
background pixel value (color), the current line style and width,
the current tile or stipple for pattern generation, the current font for text
generation, and other similar attributes.

In many graphics systems, the equivalent of the graphics context and the
drawable are combined as one entity.
The main distinction between the two kinds of status is that a drawable
describes a writing surface and the writings that may have already been done
on it, whereas a graphics context describes the drawing process.
A drawable is like a chalkboard.
A GC is like a piece of chalk.

Unlike many similar systems, there is no "current pen location."
Every graphic operation is accompanied by the coordinates where it is to happen.

The GC also includes procedure pointers
that carry out the fundamental graphic operations
such as drawing lines, polygons, arcs, text, and copying bitmaps.
The DDX graphic software can, if it
wants to be smart, change these procedure pointers
to take advantage of hardware/firmware in the server machine, which can do
a better job under certain circumstances.

The DDX software is notified any time the client (or DIX) uses a changed GC.
For instance, if the hardware has special support for drawing fixed-width fonts,
DDX can intercept changes to the current font in a GC just before drawing is done.
It can plug into either a fixed-width procedure
that makes the hardware draw characters, or a variable-width procedure that carefully
lays out glyphs by hand in software, depending upon the new font that is selected.

A definition of this block can be found in the file 
server/include/gcstruct.h.

Also included in this block is a device-private field.
DDX can load this pointer at GC creation with a pointer to
a device-private context block.

.NH 3
Details of operation
.XS
Details of operation
.XE
.LP
At screen initialization, a screen must supply a GC creation procedure.
At GC creation, the screen must fill in a GCInterestRec
(server/include/gcstruct.h) and
specify a list of entry points
to be called when that GC is modified or deleted.
The screen may also register an entry point to be called when
the GC, just prior to a drawing operation if the state
of the drawable (i.e., it's clipList) or the GC is different.
It also indicates for what state changes
it wishes to be called for modification and validation.

When a client request is processed that results in a change
to the GC, the device-independent state of the GC is updated.
This includes a record of the state that changed.
If DDX has requested notification at GC-modify time, the ChangeGC routine is called.
This is useful for graphics subsystems that are able to process
state changes in parallel with the server CPU.
DDX may opt not to take any action at GC-modify time.
This is more efficient if multiple GC-modify requests occur
between draws using a given GC.

Validation occurs at the first draw operation that specifies
the GC after that GC was modified.
If there are GC field changes that DDX has expressed interest in,
the validation procedure is called.
DDX should then update its internal state.
DDX internal state may be stored as one or more of the following:
1) device private block on the GC; 2) hardware state; 3) changes to
the GC vectors.

The GC contains a serial number, which is loaded with a number fetched from
the window that was drawn into the last time the GC was used.  The serial
number in the drawable is changed when the drawable's
clipList or absCorner changes.  Thus, by
comparing the GC serial number with the drawable serial number, DIX can
force a validate if the drawable has been changed since the last time it
was used with this GC.

In addition, the drawable serial number is always guaranteed to have the
most significant bit set to 0.  Thus, the DDX layer can set the most
significant bit of the serial number to 1 in a GC to force a validate the next time
the GC is used.  DIX also uses this technique to indicate that a change has
been made to the GC by way of a SetGC, a SetDashes or a SetClip request.

.NH 3
GC Handling Routines
.XS
GC Handling Routines
.XE
.LP
The ScreenRec data structure has a pointer for
CreateGC().
All of the rest of the routines in this section are pointed to by a GCInterestRec.
.nf

	Bool pScreen->CreateGC(pGC)
		GCPtr pGC;
.fi
This routine must fill in the fields of
a dynamically allocated GC that is passed in.
It does NOT allocate the GC record itself or fill
in the defaults; DIX does that.

This must fill in all the functions in the GC; none of the drawing
functions will be called before the GC has been validated,
but the others (dealing with allocating of clip regions,
changing and destroying the GC, etc.) might be.
It must also dynamically allocate and fill in the first GCInterestRec.
(The default font is set elsewhere; see the section on Fonts.)

The GCInterestRec is a struct used as part of a GC.
Every GC has a minimum of one GCInterestRec; they form a
doubly linked list hanging off of a GC.
Each GCInterestRec indicates some piece of code that is interested in
what happens to the GC.
The first GCInterestRec to be attached should belong to the server's DDX code.
Additional GCInterestRec's are usually the work of server extensions 
(see the extensions document for more details).
They can be attached on-the-fly, and may come before or after the server's;
they each get their own destroy routine to clean up after themselves.

Each GCInterestRec contains pointers to five
routines:
.nf

	pGCInterest->ChangeGC(pGC, pGCI, changes)
		GCPtr pGC;
		GCInterestPtr pGCI;
		Mask changes;
.fi

This GC Interest routine is called
immediately after a field in the GC is changed.
The pointer to this routine can be NULL in the GCInterestRec
if there is no routine to be called.
changes is a bit mask indicating the
changed fields of the GC in this request.
pGCI points to this GCInterestRec (handy if you need it).

The ChangeGC routine is useful if you have a system where state-changes to the
GC can be swallowed immediately by your graphics system,
and a validate is not necessary.

You can control what changes the ChangeGC routine gets called for by
setting the ChangeInterestMask in the GCInterestRec.
.nf

	pGCInterest->ValidateGC(pGC, pGCI, changes, pDraw)
		GCPtr pGC;
		GCInterestPtr pGCI;
		Mask changes;
		DrawablePtr pDraw;
.fi

ValidateGC is called by DIX
just before the GC will be used when one of many possible changes to the GC
or the graphics system
has happened. 
It can modify the devPrivate
field of the GC or its contents, change the procedure vectors,
or change hardware according to the values in the GC.
It may not
change the device-independent portion of the GC itself.

In almost all cases, your ValidateGC() procedure should take the regions 
that drawing needs to be clipped to and combine them into a composite
clip region, which you keep a pointer to in the private part of the GC.
In this way, your drawing primitive routines (and whatever is below them)
can easily determine what to clip and where.
You should combine the regions clientClip (the region that the client desires to
clip output to) and the region returned by NotClippedByChildren(), in DIX.
An example is in server/ddx/mfb/mfbgc.c.

Some kinds of extension software may cause this routine to be called more than originally
intended; you should not rely on algorithms that will break under such circumstances.

See the Strategies document for more information on creatively using 
this routine.

You can control what changes the ValidateGC routine gets called for by
setting the ValInterestMask in the GCInterestRec.
.nf

	pGCInterest->CopyGCSource(pGC, pGCI, mask, destGC)
		GCPtr pGC;
		GCInterestPtr pGCI;
		Mask mask;
		GCPtr destGC;

	pGCInterest->CopyGCDest(pGC, pGCI, mask, srcGC)
		GCPtr pGC;
		GCInterestPtr pGCI;
		Mask mask;
		GCPtr srcGC;
.fi

These routines are called by DIX 
when a GC is being copied to another GC.
After a new GC is allocated, CopyGCSource is called for the source GC;
then CopyGCDest is called for the destination GC.
This is for situations where dynamically allocated chunks of memory are hanging
off the GCInterestRec's extPriv component, and they need to be duplicated when
the GC and GCInterestRec's are duplicated.
.nf

	pGCInterest->DestroyGC(pGC, pGCI)
		GCPtr pGC;
		GCInterestPtr pGCI;
.fi
This routine is called before the GC is destroyed for the
entity interested in this GC to clean up after itself.
This routine is responsible for detaching its GCInterestRec
and freeing it, in addition to any auxiliary storage allocated
by the interested entity.

.NH 3
GC Clip Region Routines
.XS
GC Clip Region Routines
.XE
.LP
The GC clientClip field requires three procedures to manage it.
These procedures are in the GC.  The underlying principle is that
dix knows nothing about the internals of the clipping information,
(except when it has come from the client), and so calls ddX whenever
it needs to copy, set, or destroy such information.  It could have been
possible for dix not to allow ddX to touch the field in the GC,
and require it to keep its own copy in devPriv,
but since clip masks can be very large, this seems like a bad
idea.  Thus, the server allows ddX to do whatever it wants
to the clientClip field of the GC, but requires it to do all
manipulation itself.
.nf

	void pGC->ChangeClip(pGC, type, pValue, nrects)
		GCPtr pGC;
		int type;
		char *pValue;
		int nrects;
.fi
This routine is called whenever the client changes the client clip region.
The pGC points to the GC involved, the type tells what form the 
region has been sent in.
If type is CT_NONE, then there is no client clip.  If 
type is CT_UNSORTED, CT_YBANDED or CT_YXBANDED, then pValue pointer to a 
list of rectangles, nrects long.  If type is CT_REGION, then
pValue pointer to a RegionRec from the mi region code.  If type
is CT_PIXMAP pValue is a pointer to a pixmap.  (The defines
for CT_NONE, etc. are in server/include/gc.h.)
This routine is responsible for incrementing any necessary reference
counts (e.g. for a pixmap clip mask) for the new clipmask and 
freeing anything that used to be in the GC's clipMask field.  
The lists of rectangles passed in can be freed with Xfree(), the 
regions can be destroyed with the RegionDestroy field in the screen, 
and pixmaps can be destroyed by calling the screen's DestroyPixmap 
function.
DIX and MI code expect what they pass in to this to be freed
or otherwise inaccessible, and will never look inside what's
been put in the GC.  This is a good place to be wary of
storage leaks.
.LP
In the sample server, this routine transforms either the bitmap or 
the rectangle list into a region, so that future routines will 
have a more predictable 
starting point to work from.
(The validate routine must take this client clip region and merge it
with other regions to arrive at a composite clip region before any drawing
is done.)
.nf

	void pGC->DestroyClip(pGC)
		GCPtr pGC;
.fi
This routine is called whenever the client clip region must be destroyed.
The pGC points to the GC involved.  This call should set the clipType
field of the GC to CT_NONE.
In the sample server, the pointer to the client clip region is set to NULL
by this routine after destroying the region, so that other software
(including ChangeClip() above) will recognize that there is no client clip region.
.nf
	void pGC->CopyClip(pgcDst, pgcSrc)
		GCPtr pgcDst, pgcSrc;
.fi
This routine makes a copy of the clipMask and clipType from pgcSrc
into pgcDst.  It is responsible for destroying any previous clipMask
in pgcDst.  The clip mask in the source can be the same as the
clip mask in the dst (clients do the strangest things), so care must 
be taken when destroying things.  This call is required because dix
does not know how to copy the clip mask from pgcSrc.

.NH 2
Drawing Primitives
.XS
Drawing Primitives
.XE
.LP
The X protocol (rules for the byte stream that goes between client and server)
does all graphics using primitive
operations, which are called Drawing Primitives.
These include line drawing, area filling, arcs, and text drawing.
Your implementation must supply 16 routines 
to perform these on your hardware.
(The number 16 is arbitrary.)

More specifically, 16 procedure pointers are in each
GC.
At any given time, ALL of them MUST point to a valid procedure that
attempts to do the operation assigned, although
the procedure pointers may change and may
point to different procedures to carry out the same operation.
A simple server will leave them all pointing to the same 16 routines, while
a more optimized implementation will switch each from one
procedure to another, depending upon what is most optimal
for the current GC and drawable.

The sample server contains a considerable chunk of code called the
mi (machine independent)
routines, which serve as drawing primitive routines.
Many server implementations will be able to use these as-is,
because they work for arbitrary depths.
They make no assumptions about the formats of pixmaps
and frame buffers, since they call a set of routines
known as the "Pixblit Routines" (see next section).
They do assume that the way to draw is
through these low-level routines that apply pixel values rows at a time.
If your hardware or firmware gives more performance when
things are done differently, you will want to take this fact into account
and rewrite some or all of the drawing primitives to fit your needs.

.NH 3
GC Components
.XS
GC Components
.XE
.LP
This section describes the fields in the GC that affect each drawing primitive.
The only primitive that is not affected is GetImage, which does not use a GC
because its destination is a protocol-style bit image.
Since each drawing primitive mirrors exactly the X protocol request of the
same name, you should refer to the X protocol specification document
for more details.

ALL of these routines MUST CLIP to the
appropriate regions in the drawable.
Since there are many regions to clip to simultaneously, 
your ValidateGC routine should combine these into a unified 
clip region to which your drawing routines can quickly refer.
This is exactly what the mfb routines supplied with the sample server
do.
The mi implementation passes responsibility for clipping while drawing
down to the Pixblit routines.

Also, all of them must adhere to the current plane mask.
The plane mask has one bit for every bit plane in the drawable;
only planes with 1 bits in the mask are affected by any drawing operation.  

All functions except for ImageText calls must obey the alu function.
This is usually Copy, but could be any of the allowable 16 raster-ops.

All of the functions, except for CopyArea, might use the current
foreground and background pixel values.
Each pixel value is 32 bits.
These correspond to foreground and background colors, but you have
to run them through the colormap to find out what color the pixel values
represent.  Do not worry about the color, just apply the pixel value.

The routines that draw lines (PolyLine, PolySegment, PolyRect, and PolyArc)
use the line width, line style, cap style, and join style.
Line width is in pixels.
The line style specifies whether it is solid or dashed, and what kind of dash.
The cap style specifies whether Rounded, Butt, etc.
The join style specifies whether joins between joined lines are Miter, Round or Beveled.
When lines cross as part of the same polyline, they are assumed to be drawn once.
(See the X protocol specification for more details.)

Zero-width lines are NOT meant to be really zero width; this is the client's way
of telling you that you can optimize line drawing with little regard to
the end caps and joins.
They are called "thin" lines and are meant to be one pixel wide.
These are frequently done in hardware or in a streamlined assembly language
routine.

Lines with widths greater than zero, though, must all be drawn with the same
algorithm, because client software assumes that every jag on every
line at an angle will come at the same place.
Two lines that should have
one pixel in the space between them
(because of their distance apart and their widths) should have such a one-pixel line 
of space between them if drawn, regardless of angle.

The solid area fill routines (FillPolygon, PolyFillRect, PolyFillArc)
all use the fill rule, which specifies subtle interpretations of
what points are inside and what are outside of a given polygon.
The PolyFillArc routine also uses the arc mode, which specifies
whether to fill pie segments or single-edge slices of an ellipse.

The line drawing, area fill, and PolyText routines must all
apply the correct "fill style."
This can be either a solid foreground color, a transparent stipple,
an opaque stipple, or a tile.
Stipples are bitmaps where the 1 bits represent that the foreground color is written,
and 0 bits represent that either the pixel is left alone (transparent) or that
the background color is written (opaque).
A tile is a pixmap of the full depth of the GC that is applied in its full glory to all areas.
The stipple and tile patterns can be any rectangular size, although some implementations
will be faster for certain sizes such as 8x8 or 32x32.
The mi implementation passes this responsibility down to the Pixblit routines.

See the X protocol document for full details.
The description of the CreateGC request has a very good, detailed description of these
attributes.

.NH 3
The Primitives
.XS
The Primitives
.XE
.LP
The Drawing Primitives are as follows:

.nf

	RegionPtr pGC->CopyArea(src, dst, pGC, srcx, srcy, w, h, dstx, dsty)
		DrawablePtr dst, src;
		GCPtr pGC;
		int srcx, srcy, w, h, dstx, dsty;
.fi
CopyArea copies a rectangle of pixels from one drawable to another of
the same depth.  To effect scrolling, this must be able to copy from
any drawable to itself, overlapped.  No squeezing or stretching is done
because the source and destination are the same size.  However,
everything is still clipped to the clip regions of the destination
drawable.

If pGC->graphicsExposures is True, any portions of the desination which
were not valid in the source (either occluded by covering windows, or
outside the bounds of the drawable) should be collected together and
returned as a region (if this resultant region is empty, NULL can be
returned instead).  Furthermore, the invalid bits of the source are
not copied to the destination and (when the destination is a window)
are filled with the background tile.  The sample routine
miHandleExposures generates the appropriate return value and fills the
invalid area using pWindow->PaintWindowBackground.

For instance, imagine a window that is partially obscured by other
windows in front of it.  As text is scrolled on your window, the pixels
that are scrolled out from under obscuring windows will not be
available on the screen to copy to the right places, and so an exposure
event must be sent for the client to correctly repaint them.  Of
course, if you implement some sort of backing store, you could do this
without resorting to exposure events.

An example implementation is mfbCopyArea() in server/ddx/mfb/mfbbitblt.c.
.nf

	RegionPtr pGC->CopyPlane(src, dst, pGC, srcx, srcy, w, h, dstx, dsty, plane)
		DrawablePtr dst, src;
		GCPtr pGC;
		int srcx, srcy, w, h, dstx, dsty;
		long plane;
.fi
CopyPlane must copy one plane of a rectangle from the source
drawable onto the destination drawable.
Because this routine only copies one bit out of each pixel,
it can copy between drawables of different depths.
This is the only way of copying between drawables of different
depths, except for copying bitmaps to pixmaps and applying foreground
and background colors to it.
All other conditions of CopyArea apply to CopyPlane too.

An example implementation is mfbCopyPlane() in 
server/ddx/mfb/mfbbitblt.c.
.nf

	pGC->PolyPoint(dst, pGC, mode, n, pPoint)
		DrawablePtr dst;
		GCPtr pGC;
		int mode;
		int n;
		POINT *pPoint;
.fi
PolyPoint draws a set of one-pixel dots (foreground color)
at the locations given in the array.
mode is one of the defined constants Origin (absolute coordinates) or Previous
(each coordinate is relative to the last).
Note that this does not use the background color or any tiles or stipples.

Example implementations are mfbPolyPoint() in server/ddx/mfb/mfbpolypnt.c and 
miPolyPoint in server/ddx/mi/mipolypnt.c.
.nf

	pGC->Polylines(dst, pGC, mode, n, pPoint)
		DrawablePtr dst;
		GCPtr pGC;
		int mode;
		int n;
		POINT *pPoint;
.fi
Similar to PolyPoint, Polylines draws lines between the locations given in the array.
Zero-width lines are NOT meant to be really zero width; this is the client's way of 
telling you that you can maximally optimize line drawing with little regard to
the end caps and joins.
mode is one of the defined constants Previous or Origin, depending upon
whether the points are each relative to the last or are absolute.

Example implementations are the mi routines miZeroLine() in 
server/ddx/mi/mizerline.c, miWideLine() 
in server/ddx/mi/milines.c, 
miWideDash() in server/ddx/mi/miwidedash.c, and the mfb routines 
mfbLineSS() and mfbDashLine() 
in server/ddx/mfb/mfbline.c.
.nf

	pGC->PolySegment(dst, pGC, n, pPoint)
		DrawablePtr dst;
		GCPtr pGC;
		int n;
		POINT *pPoint;
.fi
PolySegments draws unconnected
lines between pairs of points in the array; the array must be of
even size; no interconnecting lines are drawn.

An example implementation is miPolySegment() in mipolyseg.c.
.nf

	pGC->PolyRectangle(dst, pGC, n, pRect)
		DrawablePtr dst;
		GCPtr pGC;
		int n;
		RECT *pRect;
.fi
PolyRectangle draws outlines of rectangles for each rectangle in the array.

An example implementation is miPolyRectangle() in server/ddx/mi/mipolyrect.c.
.nf

	pGC->PolyArc(dst, pGC, n, pArc)
		DrawablePtr dst;
		GCPtr pGC;
		int n;
		xArc*pArc;
.fi
PolyArc draws connected conic arcs according to the descriptions in the array.
See the protocol specification for more details.

An example implementation is miPolyArc() in server/ddx/mi/miarc.c.
.nf

	pGC->FillPolygon(dst, pGC, shape, mode, count, pPoint)
		DrawablePtr dst;
		GCPtr pGC;
		int shape;
		int mode;
		int count;
		POINT *pPoint;
.fi
FillPolygon fills a polygon specified by the points in the array
with the appropriate fill style.
If necessary, an extra border line is assumed between the starting and ending lines.
The shape can be used as a hint
to optimize filling; it indicates whether it is convex (all interior angles
less than 180), nonconvex (some interior angles greater than 180 but
border does not cross itself), or complex (border crosses itself).
You can choose appropriate algorithms or hardware based upon mode.
mode is one of the defined constants Previous or Origin, depending upon
whether the points are each relative to the last or are absolute.

An example implementation is miFillPolygon() in server/ddx/mi/mipoly.c.
.nf

	pGC->PolyFillRect(dst, pGC, n, pRect)
		DrawablePtr dst;
		GCPtr pGC;
		int n;
		RECT *pRect;
.fi
PolyFillRect fills multiple rectangles.

Example implementations are mfbPolyFillRect() in server/ddx/mfb/mfbfillrct.c and 
miPolyFillRect() in server/ddx/mi/mifillrct.c.
.nf

	pGC->PolyFillArc(dst, pGC, n, pArc)
		DrawablePtr dst;
		GCPtr pGC;
		int n;
		ARC *pArc;
.fi
PolyFillArc fills a shape for each arc in the
list that is bounded by the arc and one or two
line segments with the current fill style.

An example implementation is miPolyFillArc() in server/ddx/mi/miarc.c.
.nf

	pGC->PutImage(dst, pGC, depth, x, y, w, h, leftPad, format, pBinImage)
		DrawablePtr dst;
		GCPtr pGC;
		int x, y, w, h;
		int format;
		char *pBinImage;
.fi
PutImage copies a
pixmap image 
into the drawable.
The pixmap image must be in X protocol format (either Bitmap,
XYPixmap, or ZPixmap), and format tells the format.
(See the X protocol specification for details on these formats).
You must be able to accept all three formats, because the client
gets to decide which format to send.
Either the drawable and the pixmap image have the same depth, or the source 
pixmap image must be a Bitmap.
If a Bitmap, the foreground and background colors will be applied 
to the destination.

Example implementations are mfbPutImage() in server/ddx/mfb/mfbimage.c and 
miPutImage() in server/ddx/mfb/mibitblt.c.
.nf

	pScreen->GetImage(src, x, y, w, h, format, planeMask, pBinImage)
		 DrawablePtr src;
		 int x, y, w, h;
		 long format;
		 pointer pBinImage;
.fi
GetImage copies the bits from the source drawable into
the destination pointer.  The bits are written into the buffer
according to the server-defined pixmap padding rules.
pBinImage is guaranteed to be big enough to hold all
the bits that must be written.

This routine does not correspond exactly to the X protocol
GetImage request, since DIX has to break the reply up into
buffers of a size requested by the transport layer.
If format is ZPixmap, the bits are written in the ZFormat
for the depth of the drawable; if there is a 0 bit in the
planeMask for a particular plane, all pixels must have the bit
in that plane equal to 0.
If format is XYPixmap, planemask is guaranteed to have a single
bit set; the bits should be written in Bitmap format, which
is the format for a single plane of an XYPixmap.

Example implementations are mfbGetImage() in server/ddx/mfb/mfbimage.c and 
miGetImage() in server/ddx/mi/mibitblt.c.
.nf

	pGC->ImageText8(pDraw, pGC, x, y, count, chars)
		DrawablePtr pDraw;
		GCPtr pGC;
		int x, y;
		int count;
		char *chars;
.fi
ImageText8 draws text.
The text is drawn in the foreground color;
the background color fills the remainder of the character rectangles.
The coordinates specify the baseline and start of the text.

An example implementation is miImageText8() in server/ddx/mi/mipolytext.c.
.nf

	pGC->PolyText8(pDraw, pGC, x, y, count, chars)
		DrawablePtr pDraw;
		GCPtr pGC;
		int x, y;
		int count;
		char *chars;
.fi
PolyText8 works like ImageText8, except it draws with
the current fill style for special effects such as 
shaded text.
See the X protocol specification for more details.

An example implementation is miPolyText8() in server/ddx/mi/mipolytext.c.
.nf

	pGC->PolyText16(pDraw, pGC, x, y, count, chars)
		DrawablePtr pDraw;
		GCPtr pGC;
		int x, y;
		int count;
		unsigned short *chars;

	pGC->ImageText16(pDraw, pGC, x, y, count, chars)
		DrawablePtr pDraw;
		GCPtr pGC;
		int x, y;
		int count;
		unsigned short *chars;
.fi
These two routines are the same as the "8" versions,
except that they are for 16-bit character codes (useful 
for oriental writing systems).

The primary difference is in the way the character information is looked up.
The 8-bit and the 16-bit versions obviously have different kinds of character
values to look up; 
the main goal of the lookup is to provide a pointer to the CharInfo structs
for the characters to draw and to pass these pointers to the Glyph routines.
Given a CharInfo struct, lower-level software can draw the glyph desired
with little concern for other characteristics of the font.

16-bit character fonts have 
a row-and-column scheme, where the 2bytes of the 
character code constitute the row and column in a square matrix of CharInfo
structs.
Each font has row and column minimum and maximum values; the CharInfo
structures form a two-dimensional matrix.

Example implementations are miPolyText16() and 
miImageText16() in server/ddx/mi/mipolytext.c.

See the X protocol specification for more details on these graphic operations.
.LP
There is a hook in the GC, called LineHelper, that is used in the
sample implementation by the code for wide lines and arcs, both
dashed and undashed.  It draws a polyline of the proper width with
the property join style and applies the appropriate end styles if
asked.   In the sample, there are two different routines that
might get hooked in; one draws lines with mitered joints and the
other handles bevel and round joints.  
.LP
Even though this hook is stored in the GC, it is not intended to be of 
general utility.  It is only to be called by the PolyLine code.  In
specific, this routine is not intended for use by extensions.
.nf

	void pGC->LineHelper(pDrawable, pGC, docaps, nPoints, pPoints, xOrg, yOrg)
		DrawablePtr pDrawable;
		GCPtr pGC;
		Bool docaps;
		int nPoints;
		SppPointPtr *pPoints;
		int xOrg, yOrg;
.fi
The polyline consists of the nPoints points pointed to by pPoints.
(The type of nPoints is that of an SPP point; in the sample server
the coordinates are double, but you may want to optimize it
to use fixed point numbers.  See the Strategies document, section on 
optimizing, for more details.)
xOrg and yOrg are offsets that should be added to all coordinates.
docaps is TRUE if you should draw caps on the ends of the segments
and FALSE if you should not.

The LineHelper is called once for normal wide lines and once for each
segment of a wide dashed line.

The two implementations of this routine, miMiter() and miNotMiter(), are
in server/ddx/mi/milines.c.

.NH 2
Pixblit Procedures
.XS
Pixblit Procedures
.XE
.LP
The Drawing Primitive functions must be defined for your server.
One possible way to do this is to use the mi routines from the sample server.
If you choose to use the mi routines (even part of them!) you must implement
these Pixblit routines.
These routines read and write pixel values 
and deal directly with the image data.

The Pixblit routines for the sample server are part of the
"mfb" routines for Monochrome Frame Buffer.
As with the mi routines, the mfb routines
are portable but are not as portable
as the mi routines.

The mfb routines only work for monochrome frame
buffers, the simplest type of display.
Furthermore, they only work for screens that organize their
bits in rows of pixels on the screen.
(See the Strategies document for more details on porting mfb.)

In other words, if you have a "normal" 1-deep display, you can probably
use both the mfb and the mi code.
If you have a deeper frame buffer, you will have to supply
your own Pixblit routines, but you can use the mi routines
on top of them.
If you have better ways of doing some of the Drawing Primitive functions,
then you may want to supply some of your own Drawing Primitive routines.
(Even people who write their own
Drawing Primitives save at least some of the mi code for certain 
special cases that their hardware or library or fancy algorithm does not handle.)

The client, DIX, and the
machine-independent routines do not carry the final responsibility of clipping.
They all depend
upon the Pixblit routines to do their clipping for them.
The rule is, if you touch the frame buffer, you clip.

(The higher level routines may decide to clip at a high level, 
but this is only for increased performance and cannot substitute for 
bottom-level clipping.
For instance, the mi routines, DIX, or the client may decide to 
check all character strings to be drawn
and chop off all characters that would
not be displayed.  
If so, it must retain the character on the edge that is partly displayed
so that the Pixblit routines can clip off precisely at the right place.)

To make this easier, all of the reasons to clip 
can be combined into one region in your ValidateGC procedure.
You 
take this composite clip region with you into the Pixblit routines.
(The sample server does this.)

Also, FillSpans() has to apply tile and stipple patterns.
The patterns are all aligned to the window origin so that 
when two people write patches that are contiguous, they will merge 
nicely.
(Really, they are aligned to the patOrg point in the GC.
This defaults to (0, 0) but can be set by the client to anything.)

However, the mi routines can
translate (relocate) the points  from window-relative to screen-relative
if desired.
If you set the miTranslate field in the GC (set it in the CreateGC or
ValidateGC routine),
then the mi output routines will translate all coordinates.
If it is false, then the coordinates will be passed window-relative.
Screens with no hardware translation will probably set miTranslate
to TRUE, so that geometry (e.g. polygons, rectangles) can be
translated, rather than having the resulting list of scanlines
translated; this is good because the list vertices in a drawing request
will generally be much smaller than the list of scanlines it produces.
Similarly, hardware that does translation can set miTranslate to
FALSE, and avoid the extra addition per vertex, which can be (but is
not always) important for getting the highest possible performance.
(Contrast the behavior of GetSpans, which is not expected to be 
called as often, and so has different constraints.)
The miTranslate field is settable in each GC, if , for example,
you are mixing several kinds of destinations (offscreen pixmaps,
main memory pixmaps, backing store, and windows), all of whcih have
different requirements, on one screen.

As with other drawing routines, there are fields in the GC to direct
higher code to the correct routine to execute for each function.
In this way, you can optimize for special cases, for example, drawing solids
versus drawing stipples.

The Pixblit routines are broken up into three sets.
The Span routines
simply fill in rows of pixels.
The Glyph routines fill in character glyphs.
The PushPixels routine is a three-input bitblt for
more sophisticated image creation.

It turns out that the Glyph and PushPixels routines actually have a
machine-independent implementation that depends upon the Span routines.
If you are really pressed for time, you can use these 
versions, although they are quite slow.

If you use the miGetImage() code, you need to supply a
macro (or function) called GetBitFromPixel.
.nf

	GetBitFromPixel(pixel, plane, depth)
	    CARD32 pixel;
	    int plane, depth;
.fi
Pixel is a pixel value of the specified depth; this returns
the bit corresponding to plane from the pixel.  The returned value is
cast to be (unsigned char).  0 is the least significant plane, depth-1
the most significant.

.NH 3
Span Routines
.XS
Span Routines
.XE
.LP
For these routines, all graphic operations have been reduced to "spans."
A span is a horizontal row of pixels.
If you can design these routines to write into and read from
rows of pixels at a time, you can use the mi routines.

Each routine takes
a destination drawable to draw into, a GC to use while drawing,
the number of spans to do, and two pointers to arrays that indicate the list
of starting points and the list of widths of spans.
.nf

	pGC->FillSpans(dst, pGC, nSpans, pPoints, pWidths, sorted)
		DrawablePtr dst;
		GCPtr pGC;
		int nSpans;
		POINT *pPoints;
		int *pWidths;
		int sorted;
.fi
FillSpans should fill horizontal rows of pixels with
the appropriate patterns, stipples, etc.,
based on the values in the GC.
The starting points are in the array at pPoints; the widths are in pWidths.
If sorted is true, the scan lines are in increasing y order, in which case
you may be able to make assumptions and optimizations.
.LP
GC components: alu, clipOrg, clientClip, and fillStyle.
.LP
GC mode-dependent components: fgPixel (for fillStyle Solid); tile, patOrg
(for fillStyle Tile); stipple, patOrg, fgPixel (for fillStyle Stipple);
and stipple, patOrg, fgPixel and bgPixel (for fillStyle OpaqueStipple).

Example implementations are mfbSolidFS(), mfbTileFS(), mfbUnnaturalTileFS(),
mfbStippleFS() and mfbUnnaturalStippleFS(), all in server/ddx/mfb/mfbfillsp.c.
Validate routines like mfbValidateGC() switch between them.

.nf

	void pGC->SetSpans(pDrawable, pGC, pSrc, ppt, pWidths, nSpans, sorted)
		DrawablePtr pDrawable;
		GCPtr pGC;
		unsigned int *pSrc;
		POINT *pPoints;
		int *pWidths;
		int nSpans;
		int sorted;
.fi
For each span, this routine should copy pWidths bits from pSrc to
pDrawable at pPoints using the raster-op from the GC.
If sorted is true, the scan lines are in increasing y order.
The pixels in pSrc are
padded according to the screen's padding rules.
These
can be used to support
interesting extension libraries, for example, shaded primitives.   It does not
use the tile and stipple.
.LP
GC components: alu, clipOrg, and clientClip
.LP
An example implementation is mfbSetSpans() in server/ddx/mfb/mfbsetsp.c.

The above functions are expected to handle all modifiers in the current
GC.  Therefore, it is expedient to have
different routines to quickly handle common special cases
and reload the procedure pointers
at validate time, as with the other output functions.
.nf

	unsigned int *pScreen->GetSpans(pDrawable, wMax, pPoints, pWidths, nSpans)
		DrawablePtr pDrawable;
		int wMax;
		POINT *pPoints;
		int *pWidths;
		int nSpans;
.fi
For each span, GetSpans gets bits from the drawable starting at pPoints
and continuing for pWidths bits.
Each scanline returned will be server-scanline padded.
The routine can return NULL if memory cannot be allocated to hold the
result.

GetSpans never translates -- for a window, the coordinates are
already screen-relative.
Consider the case of hardware that doesn't do translation:
the mi code that calls ddX will translate each shape (rectangle,
polygon,. etc.) before scan-converting it, which requires many
fewer additions that having GetSpans translate each span does.
Conversely, consider hardware that does translate: it can set its
translation point to (0, 0) and get each span, and the only
penalty is the small number of additions required to translate each
shape being scan-converted by the calling code.
Contrast the behavior of FillSpans and SetSpans (discussed
above under miTranslate), which are expected to be used more
often.

Thus, the penalty to hardware that does hardware translation is
negligible, and code that wants to call GetSpans() is greatly
simplified, both for extensions and the machine-independent
core implementation.

An example implementation is mfbGetSpans() in server/ddx/mfb/mfbgetsp.c.

.NH 4
Glyph Routines
.XS
Glyph Routines
.XE
.LP
The Glyph routines draw individual character glyphs for text drawing requests.

You have a choice in implementing these routines.
You can use the mi versions;
they depend ultimately upon the span routines.
Although text drawing will work, it will be very slow.

If you use mfb, you can use the mfb versions of the Glyph routines.
Otherwise, you will have to write your own.
.nf

	void pGC->PolyGlyphBlt(pDrawable, pGC, x, y, nglyph, ppci, pglyphBase)
		DrawablePtr pDrawable;
		GCPtr pGC;
		int x , y;
		unsigned int nglyph;
		XCHARINFO **ppci;		/* array of character info */
		char *pglyphBase;		/* start of array of glyphs */

.fi
.LP
GC components: alu, clipOrg, clientClip, font, and fillStyle.
.LP
GC mode-dependent components: fgPixel (for fillStyle Solid); tile, patOrg
(for fillStyle Tile); stipple, patOrg, fgPixel (for fillStyle Stipple);
and stipple, patOrg, fgPixel and bgPixel (for fillStyle OpaqueStipple).
.nf

	void pGC->ImageGlyphBlt(pDrawable, pGC, x, y, nglyph, ppci, pglyphBase)
		DrawablePtr pDrawable;
		GCPtr pGC;
		int x , y;
		unsigned int nglyph;
		XCHARINFO **ppci;	/* array of character info */
		char *pglyphBase;	/* start of array of glyphs */

.fi
.LP
GC components: clipOrg, clientClip, font, fgPixel, bgPixel
.LP
These routines must copy
the glyphs defined by the bitmaps in pglyphBase and the font metrics in
ppci to the DrawablePtr, pDrawable.
The poly routine follows all fill, stipple, and tile rules.
The image routine simply blasts the glyph
onto the glyph's rectangle, in foreground
and background colors.

More precisely, the Image routine fills the character rectangle
with the background color, and then the glyph is applied in the foreground color.
The glyph can extend outside of the character rectangle.
ImageGlyph() is used for terminal emulators and informal
text purposes such as button labels.

The exact specification for the Poly routine is that the glyph is
painted with the current fill style.
The character rectangle is irrelevant for this operation.
PolyText, at a higher level, includes facilities for font changes within strings
and such; it is to be used for WYSIWYG word processing and similar systems.

Both of these routines must clip themselves to the overall clipping region.

Example implementations in mi are miPolyGlyphBlt() and 
miImageGlyphBlt() in server/ddx/mi/miglblt.c.

The mfb implementations are a bit harder to find.
The file server/ddx/mfb/mfbimggblt.c defines a routine named MFBIMAGEGLYPHBLT.
This source file is compiled twice by the Makefile, once with
MFBIMAGEGLYPHBLT defined to "mfbImageGlyphBltWhite" and 
once defined to "mfbImageGlyphBltBlack."
(The -D option of the cc command in the Makefile does this.)
In this way, one source defines both procedures.
Similarly, in the file server/ddx/mfb/mfbplygblt.c, the routine MFBPOLYGLYPHBLT
really defines three procedures simultaneously:  
"mfbPolyGlyphBltWhite," "mfbPolyGlyphBltBlack," 
and "mfbPolyGlyphBltInvert."
See all of these files for more details.

.NH 4
PushPixels routine
.XS
PushPixels routine
.XE
.LP
The PushPixels routine writes the current fill style onto the drawable
in a certain shape defined by a bitmap.  PushPixels is equivalent to
using a second stipple.  You can thing of it as pushing the fillStyle
through the stencil which is built by milines and miarcs. 
.LP
.nf
.ta 1i 3i
	Suppose the stencil is:	00111100
	and the stipple is:	10101010
	PushPixels result:	00101000
.fi
.LP
You have a choice in implementing this routine.
You can use the mi version which depends ultimately upon FillSpan().
Although it will work, it will be slow.
.LP
If you use mfb, you can use the mfb version of the PushPixels routine.
Otherwise, you will have to write your own.
.nf

	void pGC->PushPixels(pGC, pBitMap, pDrawable, dx, dy, xOrg, yOrg)
		GC *pGC;
		PixmapPtr pBitMap;
		DrawablePtr pDrawable;
		int dx, dy, xOrg, yOrg;
.fi
.LP
GC components: alu, clipOrg, clientClip, and fillStyle.
.LP
GC mode-dependent components: fgPixel (for fillStyle Solid); tile, patOrg
(for fillStyle Tile); stipple, patOrg, fgPixel (for fillStyle Stipple);
and stipple, patOrg, fgPixel and bgPixel (for fillStyle OpaqueStipple).

PushPixels applys the foreground color, tile, or stipple from 
the pGC through a stencil
onto pDrawable.  
pBitMap points to a stencil (of which we use an area dx
wide by dy high), which is oriented over the drawable at xOrg, yOrg.  
Where there is a 1 bit
in the bitmap, the destination is set according to the current
fill style.  
Where there is a 0 bit
in the bitmap, the destination is left the way it is.

This routine must clip to the overall clipping region.

Example implementations are mfbPushPixels() in server/ddx/mfb/mfbpushpxl.c and 
miPushPixels() in server/ddx/mi/mipushpxl.c.

.NH 2
Shutdown Procedures
.XS
Shutdown Procedures
.XE
.LP
.nf
	void AbortDDX()
	void ddxGiveUp()
.fi
.LP
Some hardware may require special work to be done before the server
exits so that it is not left in an intermediate state.
As explained in the OS layer, FatalError() will call AbortDDX() just
before terminating the server.  In addition, ddxGiveUp() will be
called just before terminating the server on a "clean" death,
right after calling KillServerResources().
What AbortDDX() and ddxGiveUP do is left unspecified,
only that stubs must exist in the ddx layer.
It is up to local implementors as to what they should accomplish before
termination.

.NH 3
Command Line Procedures
.XS
Command Line Procedures
.XE
.LP
.nf
	int ddxProcessArgument(argc, argv, i)
	    int argc;
	    char *argv[];
	    int i;

	void
	ddxUseMsg()
.fi
.LP
You should write these routines to deal with device-dependent command line
arguments.  The routine ddxProcessArgument() is called with the command line,
and the current index into argv; you should return zero if the argument
is not a device-dependent one, and otherwise return a count of the number
of elements of argv that are part of this one argument.  For a typical
option (e.g., "-realtime"), you should return the value one.  This
routine gets called before checks are made against device-independent
arguments, so it is possible to peek at all arguments or to override
device-independent argument processing.  You can document the
device-dependent arguments in ddxUseMsg(), which will be
called from UseMsg() after printing out the device-independent arguments.

.bp
.NH 1
Summary of Routines
.XS
Summary of Routines
.XE
.LP
This is a summary of the routines discussed in this document.
The procedure names are in alphabetical order.
The Struct is the structure it is attached to; if blank, this 
procedure is not attached to a struct and must be named as shown.
The sample server provides implementations in the following
categories.  Notice that many of the graphics routines have both
mi and mfb implementations.
.TS
l l.
dix	portable to all systems; do not attempt to rewrite (server/dix)
4.2	written for 4.2/4.3bsd Unix or equivalent (server/os/4.2bsd)
mfb	routine provided in server/ddx/mfb
mi	routine provided in server/ddx/mi
snf	routine provided in server/ddx/snf
hd	hardware dependent (example in server/ddx/dec/qvss)
none	not implemented in sample implementation
.TE
.TS
expand;
c c c 
l c l.
Procedure	Port	Struct
_
ALLOCATE_LOCAL	4.2
AbortDDX	hd
AddEnabledDevice	4.2
AddInputDevice	dix
AddScreen	dix
Bell	hd	Device
ChangeClip	mi	GC
ChangeGC	mfb	GCInterest
ChangeWindowAttributes	mfb	Screen
ClearToBackground	mfb	Window
ClientAuthorized	4.2
CloseFont	snf
CloseScreen	hd
ConstrainCursor	hd	Screen
CopyArea	mi, mfb	GC
CopyGCDest	mfb	GCInterest
CopyGCSource	none	GCInterest
CopyPlane	mi,mfb	GC
CopyWindow	mfb	Window
CreateGC	mfb	Screen
CreatePixmap	mfb	Screen
CreateWellKnowSockets	4.2
CreateWindow	mfb	Screen
CursorLimits	hd	Screen
DEALLOCATE_LOCAL	4.2
DescribeFont	snf
DestroyClip	mi	GC
DestroyGC	mfb	GCInterest
DestroyPixmap	mfb	Screen
DestroyWindow	mfb	Screen
DisplayCursor	hd	Screen
Error	4.2
ErrorF	4.2
ExpandFontName	4.2
ExpandFontNamePattern	4.2
FatalError	4.2
FiClose	4.2
FillPolygon	mi	GC
FillSpans	mfb	GC
FiOpenForRead	4.2
.TE
.bp
.TS
expand;
c c c 
l c l.
Procedure	Port	Struct
_
FiRead	4.2
FlushAllOutput	4.2
FlushIfCriticalOutputPending	4.2
FreeFontRecord	4.2
GetFontPath	4.2
GetGlyphs	snf
GetImage	mi,mfb	GC
GetMotionEvents	hd	Device
GetSpans	mfb	GC
GetStaticColormap	mfb	Screen
ImageGlyphBlt	mi,mfb	GC
ImageText16	mi	GC
ImageText8	mi	GC
InitInput	hd	
InitKeyboardDeviceStruct	dix	
InitOutput	hd	
InitPointerDeviceStruct	dix	
InsertFakeRequest	4.2
InstallColormap	mfb	Screen
Intersect	mi	Screen
Inverse	mi	Screen
LegalModifier	hd
LineHelper	mi	GC
ListInstalledColormaps	mfb	Screen
LookupKeyboardDevice	dix	
LookupPointerDevice	dix	
NextAvailableClient	dix
OpenFont	snf
OsInit	4.2	
PaintWindowBackground	mi,mfb	Window
PaintWindowBorder	mi,mfb	Window
PointerNonInterestBox	hd	Screen
PointInRegion	mi	Screen
PolyArc	mi	GC
PolyFillArc	mi	GC
PolyFillRect	mi,mfb	GC
PolyGlyphBlt	mi,mfb	GC
Polylines	mi,mfb	GC
PolyPoint	mi,mfb	GC
PolyRectangle	mi	GC
PolySegment	mi	GC
PolyText16	mi	GC
PolyText8	mi	GC
PositionWindow	mfb	Screen
ProcessInputEvents	hd	
PushPixels	mi,mfb	GC
PutImage	mi,mfb	GC
QueryBestSize	hd	Screen
QueryFont	snf
QueryGlyphExtents	snf
QueryTextExtents	snf
ReadRequestFromClient	4.2	
RealizeCursor	hd	Screen
.TE
.bp
.TS
expand;
c c c 
l c l.
Procedure	Port	Struct
_
RealizeFont	mfb	Screen
RealizeWindow	mfb	Screen
RecolorCursor	hd	Screen
RectIn	mi	Screen
RegionCopy	mi	Screen
RegionCreate	mi	Screen
RegionDestroy	mi	Screen
RegionEmpty	mi	Screen
RegionExtents	mi	Screen
RegionNotEmpty	mi	Screen
RegionReset	mi	Screen
ResolveColor	mfb	Screen
RegisterKeyboardDevice	dix	
RegisterPointerDevice	dix	
RemoveEnabledDevice	4.2
ResetCurrentRequest	4.2
RestoreAreas	none	BackingStore
SaveDoomedAreas	none	BackingStore
SaveScreen	mfb	Screen
SetCriticalOutputPending	4.2
SetCursorPosition	hd	Screen
SetDefaultFont	snf
SetDefaultFontPath	4.2
SetFontPath	4.2
SetInputCheck	dix	
SetSpans	mfb	GC
StoreColors	mfb	Screen
Subtract	mi	Screen
TimeSinceLastInputEvent	hd
TranslateBackingStore	none	BackingStore
TranslateRegion	mi	Screen
UninstallColormap	mfb	Screen
Union	mi	Screen
UnrealizeCursor	hd	Screen
UnrealizeFont	mfb	Screen
UnrealizeWindow	mfb	Screen
ValidateGC	mfb	GCInterest
ValidateTree	mi	Screen
WaitForSomething	4.2
WindowExposures	mi	Window
WriteToClient	4.2	
Xalloc	4.2
Xfree	4.2
Xrealloc	4.2
.TE

.TC