Age | Commit message (Collapse) | Author | Files | Lines |
|
Change-Id: I87e63008fe7c6db62c18bf461dc4dcda733393ac
|
|
Change-Id: I510e0d601e293ed8e013a0273d5ae0f9be078d8a
|
|
Change-Id: I00f3ed059bd633e662e5bee09703ca505bf3b9a5
|
|
Change-Id: I277fbf522b16d9b479260df8372a5ee160adcc37
|
|
Change-Id: I924a212210703b0f6136ddefacd77d98dc89f42d
|
|
Change-Id: I6b4a7f8053adea6d86ee625201eeead188bbadcc
|
|
Change-Id: I6a6e702e243b389f1c487b6b5390f28a4292cc74
|
|
Change-Id: Ic020d9604814213e13c339b07b6e74de77a9f400
|
|
Change-Id: I2f7967ee20ad71b58e2b0dc7f182769499910373
|
|
Change-Id: Ia2ed94dbbb43d2e768da5af5ca7a51f5cda5bae0
|
|
Using direct ByteBuffer is much nicer option to store or send
pointers between C(++) code and Java via JNI as it handles endiness
and pointer size for us. Using "long" type can have unexpected
results in 32-bit architectures (mostly Android). This was causing
grief especially when Android introduced support for 64-bit
architectures starting with SDK 19.
Change-Id: Ie92d0f913b668e1724e846d70d1820445d9cb086
|
|
Change-Id: I7688968e92ae4b3d6fef19c4544ae91bdcd8b1b9
|
|
Change-Id: Ib003da5c7fec7fc3783f01a33a63deb384c7e770
|
|
Change-Id: Ifc7b18e173b0c91c24a53fad9c35ac3a34a4b33e
|
|
Change-Id: Id90680d3b207973b55927add1c91111268bb2a48
|
|
Change-Id: Iefab77e543606cfad937a79743fb3b9a68a0f2a2
|
|
Change-Id: Ia8516da556b3736f34b366e2eb89ad8bbd7bafc1
|
|
DirectBufferAllocator is responsible to allocate buffer. We used
Fennec JNI based allocation (and freeing) because of overallocation
bug in some Android versions. With this the VM based allocator
implementation is added and used by default because of bugs that
happen in newer Android versions (specifically when ART is used
as the Android VM).
Change-Id: I07eb364fd1647b3a09d1568d4fef82398a02dfeb
|
|
Change-Id: Ied2687d334f7d1ff9ff2f3cb6664848d50814b7c
|
|
Change-Id: Ie172f60a7134173462ff9711a40dc74c94ad8206
|
|
Change-Id: If85c6a86434c0aa32d188a0f128f6b6cd613bbb5
|
|
Change-Id: I1b1891f0d7d7b8aa407e7da346ff5f8e3cbe8657
|
|
It apparently contains color schemes and stuff that the code wants access to
here and there.
Change-Id: Ie3f0de5e3846df99a02a2693156679cc6c010d10
|
|
Change-Id: I9e5bff735b0035c147e10ae066da3a4873d66749
|
|
We are experimenting with a pruned configuration database in the desktop case,
and letting the exception propagate here killed the document loading.
Change-Id: I59e5d016617c17c2bc36de2fd69c6691bfa6b135
|
|
Change-Id: Ib9cf1a47eb20bd28d954ddcded89f67cf6865f1c
|
|
This reverts commit 00d2482326af17ac61d3c2aaa21d837f0e72bb37.
|
|
Change-Id: I379601099bda928b9eeeaeb29030bc009e3cbbf2
|
|
Change-Id: I3d0db329311e9f2496f80931e6d2776e18f1b2d9
|
|
Change-Id: I83a395c628b00b6ca96c93d5d5362ea750e57b2b
|
|
Change-Id: I290e2c84b17b9b5063139c6027b72f6cd3a78a99
|
|
Change-Id: I1e71b8a6348301cd5b3fed0ae8b3346ae3e07d8e
|
|
Change-Id: I896c9ac1c941b85d052fbefb902c4341664881d4
|
|
Change-Id: I014d70ace7ce34b804ea2a018d3de8f94f7e0cbc
|
|
Change-Id: Ie6347ed6b9c1b8bee7a08950d5c67d4a2039fba5
|
|
Change-Id: I4d819f59ecb2462aee5998628ad034657bf3b0f9
|
|
Change-Id: I43c7725b88c9326e269abc277f42e47c08acefb5
|
|
Change-Id: Iaada020e85a8d2557e270f7d1514f2260ffb2af0
|
|
This was some staggering proportion of tiled rendering documents
with complex clipping; it seems 'clear' is not what memset is for
1bit clip masks.
Change-Id: I9142ffb7d7016603feb7782d6f03b9992b9494e3
|
|
The "bootstrap" errors tend to end up being displayed in some error dialog
attempt which of course does nothing on mobile platforms, and that crack seems
to crash on Android even, so this should give us some chance to get some idea
at least what the code thinks is wrong...
Change-Id: I070b017baf042ff692766d1efd1511e1addb6ecf
|
|
Change-Id: I3e3162474db1edef3b92e252b249fe373dedbbfd
|
|
Change-Id: I0da75df61d129aaaf03ac546fcf1e5dbcff9d404
|
|
We use arbitrary tags when logging stuff in our code so we can't use the
built-in filtering of adb logcat.
Change-Id: I2d607b86bde975c5cbdd17adc22d0fc15076be51
|
|
Does not work, though, we end up with a crash that is hard to debug thanks to
the rubbish tool-chain.
Change-Id: Ie1954e35e649fac8dd106f0ccbc6951c4a6c1c63
|
|
As we call AttachCurrentThread() in the constructor, it seems logical to call
DetachCurrentThread() in the destructor. Some warning messages in the logcat
indicated that DetachCurrentThread() hadn't been called for a thread that
died.
Actually I would prefer to call both AttachCurrentThread() and
DetachCurrentThread() in the actual thread function, instead of somewhat
haphazardly and imlicitly in the AndroidSalInstance constructor and
destructor. Do we know for sure that the lifetime of the AndroidSalInstance
singleton (is is a singleton, right?) is 1:1 coupled to that of the LO "main
thread"?
Change-Id: Ifcc0e0b9af88b19389c416c5646499d44ad0e941
|
|
Change-Id: I040bd49f8cfeec74c3225135a110140c1816be43
|
|
I assume, for lo_destroy() to be useful at all. Not that I know for sure what
the exact intended semantics of lo_destroy() and lo_initialize() are. But I
assume that the idea is that after lo_destroy() has been called,
lo_initialize() can be called again.
Change-Id: I2dea26db150d19ed438427e1c119a5297b9bc9c3
|
|
It is invalid in case lo_destroy() has been called.
Change-Id: I45533b66d32fc650e48748da8ea1d2f2aaa381e0
|
|
Change-Id: I8f91e73fe5df5dfef054df80d43be3c74d01388b
|
|
Otherwise the gdbserver ends up with
run-as: exec failed for /data/data/org.libreoffice/lib/gdbserver Error:Permission denied
(you need to run ndk-gdb with --verbose to see that).
Change-Id: Iccdf0ff268c20d2fb5abc1e93404375fa51c1cf1
|