Age | Commit message (Collapse) | Author | Files | Lines |
|
On a machine with MSVC 2013 and 2015 (but no 2017) installed, I installed
"Windows Software Development Kit - Windows 10.0.15063.137" as available through
<https://developer.microsoft.com/en-US/windows/downloads/windows-10-sdk> (in the
hopes that it will provide a better fix for
acbb7e3243afec515bf98b2b34454afc5320bf66 "Work around clang-cl issue with MSVC's
safeint.h for now"). That apparently removed various files from the
C:\Program Files (x86)\Windows Kits\10\bin\ hierarchy, while it added files to a
new C:\Program Files (x86)\Windows Kits\10\bin\10.0.15063.0\ hierarchy (and
<https://developer.microsoft.com/en-US/windows/downloads/windows-10-sdk>
mentions that in an "SDK Layout Changes: Side by Side install of tools and
metadata" section). So return a new winsdkbinsubdir var from find_winsdk() to
cover installations that use this new layout scheme.
In b076e8f673d1714ee8b6f991eb2ca299b6fa8949 "Fix include and library paths for
at least me", tml had hardcoded some 10.0.15063 -> 10.0.14393 mappings that are
unhelpful at least in my scenario. I tried to keep those hacks in in a way that
would still work for tml's scenario.
The check for msiinfo.exe etc. had been broken since
ecfabb66f421ef07f52a655117f047aa75d54a01 "add support for Windows 8 SDK", but
apparently in a way so that nobody noticed. The first
# needed for msi packaging
pathmunge "$WINDOWS_SDK_BINDIR_NO_ARCH/x86" "before"
is indeed needed, as (at least some of) those tools are only available in the
arm and x86 sub-dirs, not in the arm64 and x64 ones. (I'm not sure whether the
two later pathmunges adding $WINDOWS_SDK_BINDIR_NO_ARCH/{x64,x86} are still
relevant, though.)
Change-Id: Ia403e4bde7b1abf7e2c9d9f084a05af2d1311617
Reviewed-on: https://gerrit.libreoffice.org/36455
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
|
|
Change-Id: Idd192405810d1e56ded4a42a1bcf972d2e3de904
|
|
Despite that the HKLM/SOFTWARE/Microsoft/Microsoft SDKs/Windows/
v10.0/ProductVersion value is 10.0.15063, the headers and libraries we
want are in the subdirectories called 10.0.14393.0, not 10.0.15063.0.
Not sure if this is just because of the collection of Visual Studio
versions I happen to have installed, and their order of installation,
or always true.
Instead of all the convoluted logic in configure.ac, it would make
much more sense to just use the relevant vcvars64.bat or vcvars32.bat
file and pick out the INCLUDE and LIB environment variables it sets,
and mangle that. We already have a (misleadingly named) function to
look for variables set by a .bat file, win_get_env_from_vsvars32bat().
Change-Id: I647ec1d0924ba4c743e4c16d2fa0a05014e0b2e3
|
|
Unfortunately -Og doesn't work as well as advertised, variables are
optimized away too often.
See thread at
https://lists.freedesktop.org/archives/libreoffice/2017-April/077479.html
Change-Id: I5fc141ea9c7c6931aaf8220c7abf6b413326049e
|
|
added 10.3 SDK to configure.ac
Change-Id: I2bd287cada37e595ee25a5a1bfc535daeeb247a9
|
|
Change-Id: Ie244411dd155af245c83e758f432237ce72fa623
Reviewed-on: https://gerrit.libreoffice.org/35981
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: David Tardon <dtardon@redhat.com>
|
|
Change-Id: Iea46a5d41a83edb150573102ccfbd9cd3ab4e89a
Reviewed-on: https://gerrit.libreoffice.org/35980
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: David Tardon <dtardon@redhat.com>
|
|
Switch to using sha256sum for checking if files change. Not for
security, just so we don't need to check for md5sum.
We also change the Windows installer to rely on the perm md5
digest instead of the environment variable. The code to do this was
already in directory.pm
Change-Id: I24aed542c6201abf030fdd62116aec3f8ea3513b
Reviewed-on: https://gerrit.libreoffice.org/35140
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
|
|
Change-Id: Icd1ce144733d2f5e907a19bb6ef307d50becb49d
Reviewed-on: https://gerrit.libreoffice.org/35769
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: David Tardon <dtardon@redhat.com>
|
|
Change-Id: I56e429ff1b8ee55ccb8df16002b509770762172c
Reviewed-on: https://gerrit.libreoffice.org/35618
Reviewed-by: Julien Nabet <serval2412@yahoo.fr>
Tested-by: Julien Nabet <serval2412@yahoo.fr>
|
|
This was added because CPPUNITTRACE='"$(DEVENV)" /debugexe' is less
convenient to write than CPPUNITTRACE='$(DEVENV) /debugexe'.
However DavidO points out that CPPUNITTRACE=TRUE is even more
convenient and sort of optimal already, and it turns out it's broken by
the extra quotes, so remove the hack.
Change-Id: Ic5d94f9a0774cd41821b68728404c205552d674e
Reviewed-on: https://gerrit.libreoffice.org/35589
Reviewed-by: David Ostrovsky <david@ostrovsky.org>
Tested-by: Jenkins <ci@libreoffice.org>
|
|
... from a DOS style 8.3 path, for whatever reason. Special case DEVENV
variable so it uses long path.
Change-Id: I03bb20a8f35cd116edd33ec91178f9deddbc4257
|
|
...which appears to happily accept complex C++14-style constexpr function
definitions, and only complains if such a function call is then used in a
constant expression context.
Change-Id: I8f7efd7f0af7b1b98d78840cebfeda8bc96e15d0
|
|
Change-Id: I13bf9ff8e824f69f54158af1aaf6d4cb5709b04a
|
|
Change-Id: Id8985a6049f8ffe7494d04c535f981767ffc4456
|
|
prefer having nasm/yasm but if its not there warn and fallback
to disabling assembler optional goodness
Change-Id: Ib31ad81717842f743c2910d575a9ebbc279a9c79
Reviewed-on: https://gerrit.libreoffice.org/35189
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
|
|
Change-Id: I7ac4f32cb01bfcc3a2c06e69280c2150d408ec6b
Reviewed-on: https://gerrit.libreoffice.org/35174
Reviewed-by: David Ostrovsky <david@ostrovsky.org>
Reviewed-by: Tor Lillqvist <tml@collabora.com>
Tested-by: Tor Lillqvist <tml@collabora.com>
|
|
Visual Studio 2015 has a Community Edition. Use that. VS2017 has no
Express variant at all.
Change-Id: I23bcd10c543a7626b2d69d64306fa5eac74e3b27
Reviewed-on: https://gerrit.libreoffice.org/35009
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: David Ostrovsky <david@ostrovsky.org>
|
|
Change-Id: Id3723d0242711ba9b5450bb69b2ea09e614c90f6
Reviewed-on: https://gerrit.libreoffice.org/35014
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
|
|
This reverts commit 754854674859560506b617ebf07e6563c197aac8.
Change-Id: I0863fb5e34608911c784bbdf19e76297b3c2d13e
|
|
Change-Id: Iffbfc75772219cc27b28013c35c20e98cd3cacc7
|
|
Converts Windows to Unix newlines, so bash will handle the
result correctly.
Since $(..) or `..` will already remove trailing newlines,
this reverts the unneeded changes of commit
37baabe6436037de604bfbc2702bfb0bc156c41b
Change-Id: I0071237f0600c2d71da066eb2e2b541e338e7016
|
|
Change-Id: Id98d73eef46d63686745bf6cd3482b18c7223c3a
|
|
Change-Id: I7ae99ebc693291709edc6df233f54dfe00d74e9e
|
|
Change-Id: Ida91e0ec55f2bacad036bf016f7dac8c5784cc2d
|
|
Change-Id: I311f68b32bb33a7d9363c1c28dfd0e5ad8ee986c
|
|
Change-Id: Ia3a634362e1c17c5aed14fdba7396c5319110b34
|
|
The brackets already act as quotes in .ac files. Having additional
quotes causes the result to look like:
checking whether we want to fetch tarballs... "yes, if we use them"
which is not consistent.
Change-Id: Icc5dfc4c6b58d1d7a3d4bb280cc3da3dc21e515a
|
|
Change-Id: Idaad0d2d51f533100ae3d57eb933b6d1becb4f56
|
|
Use the existing convention for how the output from AC_MSG_* lines
should look.
Change-Id: Id416ac97d65aa4a7445242c9a4853b6941ae2b4e
|
|
As the patterns in the cases don't overlap, it makes no differentce in
what order they are. So keep them sorted for clarity.
Change-Id: I60f8304d66f6925028f86a2956f1136547dc51eb
|
|
The find_winsdk_version function does not accept it.
Change-Id: I1d6a5b528e9bc1c93cc589fe5832150e3994ada5
|
|
Has always been like that. Apparently I got interrupted when typing
it. No idea what I wanted to say.
Change-Id: I86ee8ac2fc682d7284a1e51c91e2e4fb37a14c10
|
|
Output just one AC_MSG_RESULT for each AC_MSG_CHECKING.
Change-Id: I2ff1458a8b4abbe1804e33a39a9f4cc795cb28a1
|
|
You are supposed to output just one AC_MSG_RESULT for each AC_MSG_CHECKING
find_ucrt already errors out if it doesn't.
Change-Id: I63f58996f60577301631b11dd51728fdee324095
|
|
The exact same shell if statement was duplicated.
Change-Id: Icc95f8a881605e6ac7248252c545e433b936f4c1
|
|
Simply strip all newlines, returns and '\0' bytes using tr and
don't append a new newline when echo'ing the extracted value.
Change-Id: Id36293042c9071d5ef664a82ebd32de287574f31
Reviewed-on: https://gerrit.libreoffice.org/34970
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
|
|
Reading Windows registry entries via "/proc/registry*" lets current
bash complain about "command substitution: ignored null byte in input".
This happens, because Cygwin '\0'-terminates all registry values.
To quieten bash, directly strip the registry values of '\0' using tr.
Change-Id: Ibb38761587691ef20911f025c711553426b1a222
Reviewed-on: https://gerrit.libreoffice.org/34969
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
|
|
Seems that they added an "onecore" directory level before the "x64".
At least in the Community Edition.
|
|
This reverts commit fb4467b6ffc6d77ab79e11806f00cf30f33388fd.
|
|
Change-Id: Idb14cdbb3f1717e5d0aaa6fddbe39d2f46384d08
|
|
Change-Id: I0e6e01918f58ada6a58f1d3b3278b72944cf300e
Reviewed-on: https://gerrit.libreoffice.org/34904
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: David Ostrovsky <david@ostrovsky.org>
|
|
This leaves all the md5sum parts in place for now, but switches
to using sha256sum. Tries both shasum -a 256 and sha256sum.
Change-Id: I393d0dde56521f6e9e76f3d5a8d15c6c830fd683
Reviewed-on: https://gerrit.libreoffice.org/34633
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>
|
|
...after 84b36c704d73362d4d86dc9e9c0efa0625958347 "Drop support for MSVC 2013".
Make this a fatal configuration error for now. The check should be removed
completely after LO 5.4 branch-off.
Change-Id: If2f196abb93607dde9ba5c4f04d219679585e633
Reviewed-on: https://gerrit.libreoffice.org/34822
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
|
|
...after 84b36c704d73362d4d86dc9e9c0efa0625958347 "Drop support for MSVC 2013".
Make this a fatal configuration error for now. The check should be removed
completely after LO 5.4 branch-off.
Change-Id: I990fd8fcb4ec1327282df4efe21640c938d3cf06
Reviewed-on: https://gerrit.libreoffice.org/34821
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
Tested-by: Stephan Bergmann <sbergman@redhat.com>
|
|
...after 5.3 branch-off, cf. 1ea987f23a4514bcae5a6bd98014ade1af4a0107 "Remove
HAVE_GCC_VISIBILITY_BROKEN, never true"
Change-Id: Ibc6a059d5704c3ceafffef578bf61abe202dbdd6
|
|
Change-Id: Ibf47c9ff7b5fb098e284a58c547b61286264dd80
Reviewed-on: https://gerrit.libreoffice.org/22588
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
|
|
The license should not be an issue any more since liblangtag now has an
LGPL/MPLv2 dual licence.
Change-Id: I4dfffcccf5d710bd16b21c3254c1449ae041b8ab
Reviewed-on: https://gerrit.libreoffice.org/34511
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Eike Rathke <erack@redhat.com>
|
|
Change-Id: Ib5d84e67d36e41d20e137ce96ce66f758b110198
Reviewed-on: https://gerrit.libreoffice.org/34596
Reviewed-by: Michael Stahl <mstahl@redhat.com>
Tested-by: Michael Stahl <mstahl@redhat.com>
|
|
Change-Id: If6cfe913445e8855eab3736d9abc63d7e152b361
|