Age | Commit message (Collapse) | Author |
|
|
|
(always-installed packages), looks like they are needed...
|
|
So, after hours of wild hunt, I think both CMake and Scons builds work again.
There were several small issues hiding here and there, esp. for scons.
*WARNING* the --all-static is currently officially broken (it was already quite useless in fact).
I’m seriously considering dropping it completely, having a script compiling a chain
of static libs in a complete working way for *two* different build systems is just
too much work (static libs are *really* a nightmare at link time).
Builders who want to generate static Blender (to share it e.g.) will have to handle things by hand for now.
Main other points changed:
* Many changes to generated settings for scons (notably, I had to hack around the fact that the short ld -l
option does not work with lib names like 'Iex-2_1', which is not that surprising).
* Opencollada remains the only lib built statically (can't get it to build dyn currently).
* All other libs, even ILMBase ones, are now fully dynamic ones.
* Some general cleanup with CMake definitions of built libs.
* Git-libs do not clone again and again the repo anymore!
|
|
|
|
|
|
Note that I also updated OpenEXR to 2.1.0 (and updated our CMake openexr module as well).
Users will have to update their build settings as told by the script (also in BUILD_NOTES.txt generated file)!
Note I removed all ugly patches from this script, we are currently using custom git repo for OpenEXR/OIIO/OSL,
hopefully we will soon be able to use vanillia code from official repos!
|
|
leave env set for OSX exclusive for now”
|
|
This is needed to resolve the real used compiler from the compatibility shim’S or symlinks.
Also set the real CCVERSION as scons CCVERSION would only read the default one, which
is not useful for OSX
|
|
submodule ), fix a compiler warning with clang < 5”
|
|
While its possible to have different C/C++ compilers, we dont support
this at the moment, so just check C compiler ID for now.
|
|
Better just use "strip", to be sure \r\n is also stripped.
This also corresponds to what happens with popen() results.
|
|
|
|
|
|
Before this console output contained messages about branch is not
configured to use any upstream branch which is not useful information
at all.
|
|
It's not needed, this should happen automatically already, and if you build
against libc++ instead (as on recent FreeBSD), this would give build errors.
Ref T37477.
|
|
Seems to be left by accident in rB9a79e03752509997e99094763f9183e6e5ce3637
If not, please do prints more verbose :)
|
|
|
|
Ok, this is nasty. SOURCE_DIR is somehow being defined
from CMakeLists.txt in source/creator by the looks of it
and actual issue as extra slash in the end of path which
was passing to EXISTS command.
|
|
- It was using SOURCE_DIR which is undefined on windows.
No idea why it was defined on Linux, could be some
co-incident. Correct usage is CMAKE_SOURCE_DIR.
- Remove usages of SVN's revision.
|
|
Before this hash of latest local commit was used,
which is not so much useful since nobody can access
that commit hash.
|
|
This is an attempt to make it so revision from git
SHA is being displayed, not from libs svn revision.
|
|
|
|
- Use -M suffix if working tree does have uncommitted
modifications.
- Local commits are considered local changes as well
|
|
Consider this configuration final-1 for the migration,
Linux and OSX slaves seems to be fine, Windows slave
would need some changes from the slave side.
|
|
|
|
Need this so checkout of existing svn does not
interferes with new git checkout which would
need to be done.
|
|
Pretty much sure more tweaks would be needed,
but need this to get started.
|
|
over .mo files, faster bundling
|
|
|
|
|
|
Fallback to "no buildinfo" mode in this case.
|
|
- Better to set branch/hash to "unknown" so
it's clear build is done from exported tree.
- SCons used to reference undefined variable.
|
|
Title says it all :)
Pointer by IRIE Shinsuke in the ML, thanks!
|
|
Summary:
Old idea with changes since previous release tag
didn't work good enough. In most of the cases tag
was done in a branch hence not actually reachable
from the master branch.
Now change since release is gone, and date of
the latest commit is used instead.
The date is displayed in format YYYY-MM-DD HH:mm
in the splash.
New bpy.app fields:
- build_commit_timestamp is an unix timestamp of
the commit blender was build from.
- build_commit_date is a date of that commit.
- build_commit_time is a time of that commit.
Reviewers: campbellbarton
Differential Revision: http://developer.blender.org/D5
|
|
Left it there by an accident.
|
|
Made it so if there's release/datafiles/locale/po
folder, then all the .po files will be converted
to .mo at blender compile time and installed to
an appropriate location.
Uses small own implementation msgfmt which is
based on msgfmt.py from Python project, but also
supports contexts.
There's no functional changes for until we've
switched to use source .po files instead of
pre-compiled .mo.
P.S. Well, there's one change which is msgfmt.cc
being compiled even if it's not used, but
would rather not clutter code with checks
since pretty soon we'll use this program
anyway.
|
|
env, also fix a longstanding misuse of MACOSX_SDK_CHECK in ceres, use MACOSX_SDK instead
|
|
|
|
|
|
|
|
|
|
desired variables changed
|
|
|
|
|
|
|
|
MACOSX_DEPLOYMENT_TARGET instead
|
|
to not need to cp whole config as user-config later
|
|
use sw_vers -productVersion instead of uname -r, we must not redine it then from darwin version -> osx version
|
|
backend.
This means that if you have WITH_BF_QUICKTIME or WITH_CODEC_QUICKTIME enabled,
it will always use QTKit.
The old backend was only used on 32 bit OS X builds, now 32 and 64 bit builds will
give consistent input/output. On Windows or Linux quicktime isn't being used.
|
|
|