we don't really care what the error number is, so long as its an error of some
kind so just sed JERR_BAD_CROP_SPEC to JERR_CONVERSION_NOTIMPL
Change-Id: Iae41de7b720ce2e60c7092d15983aa1c4ba14c89
Forcing -stdlib=c++11 into CXX upon --enable-libc++ instead of only into
CXXFLAGS_CXX11 as done normally caused external projects to unexpectedly pick it
up, causing build failures in firebird.
Change-Id: I588d75ccbde6b9fd4f38e10a0c67c2f6086f8977
Pass the information via config_host/config_buildid.h. Then
cui/source/dialogs/about.cxx will get rebuilt also in the incremental build.
Change-Id: I058cbe162e7015cbea6c71253c20db097477708f
It allows to build only the common parts of the help, e.g.
bundle the helpcontent-related icons.
It is useful when the build content is built separately. For example,
it is used to speed up the build of the main package in openSUSE
Build Service.
The installset structure in a "traditional" (non-"Mac-like") OS X was
messed up with *both* 'MacOS' and 'program' as real subfolders in
instdir.
Change-Id: I337e1669d28c6c59993790fd33c9c5ab372c421c
Lots of stuff still either ended up in the wrong place, or was looked up from
the wrong place, or both. Fix most cases.
Change-Id: I06ebbce207c219f3cd82b4387dd9b3fdb83420d4
Add a switch --enable-canonical-installation-tree-structure. Only implemented
for OS X for now, but the same switch can be used if and when we want to make
it possible to make the Unix installation tree more "standard", too.
Change-Id: Icb0ad500b77d7e986f2f188fd17faabaf2e2fe99
Change all instances of hardcoded "program", "share" etc subfolder names to
use those from <config_folders.h> instead. In normal builds, the end result
will not change.
Change-Id: I91c95cd8e482818be67307e889ae6df887763f53
First cut was at https://gerrit.libreoffice.org/#/c/2310/ but was
abandoned by myself. Now getting back to the idea, but this time I
will make it optional on OS X, and introduce it in smaller steps.
Change-Id: Id88517fee8d49056b8bbecb0921206dc6b29580b
If what we want is to compile the OpenCL code by default always on OS
X when building against a SDK newer than 10.6, that can be said
simpler.
Change-Id: I14b2cd107fe0dba42c221bec63fd4076fffbd848
Clang reports itself as GCC 4.2.1 when asked the GCC way, which is about
as good an answer as any, since there's no good mapping between them.
So when testing GCC version, Clang would be usually considered too old,
and therefore the proper way is to do a configure check if possible.
GCC version should be only used for GCC-specific things such as avoiding
a bug in a specific GCC version, and such testing should first check
the compiler is not Clang.
Rename GCCVER to GCC_VERSION and use it throughout the build system
where needed. As it's empty for anything that's not GCC, this should
lead to errors when used incorrectly.
Change-Id: Iea96bbaf5d8ceabefa25be88576eeb4115384937
Give up and revert to "$CC -dumpversion".
Since by now no test now refers to patchlevel, make it unavailable in GCCVER so that nobody tries to use it.
Change-Id: Ife0bb0a94a73555d1f19f5a3c5853148042c7bba
...so extend 409e6a2e0b89eab82e79b0cd122062f47c6c7afb "gcc -dumpversion has only
MAJOR.MINOR, not MAJOR.MINOR.PATCHLEVEL" to also cope with e.g. Fedora 18 gcc's
"gcc (GCC) 4.7.2 20121109 (Red Hat 4.7.2-8)" output line.
Change-Id: Ieb5e032685da49cd7e6f0a49a0fdc137fc24fbbd
At least, for gcc 4.7 on Debian. OTOH, gcc --version outputs
(on its first line)
gcc-4.7.real (Debian 4.7.2-5) 4.7.2
so use that. Hopefully will work across systems & distros.
(gcc-4.7.real is probably just argv[0],
because diversion from hardening-wrapper)
Change-Id: I94ea73628d8aac107e3ada1dc5a63e51e79ccdb2
