| Age | Commit message (Collapse) | Author |
|---|
|
|
|
gb_LinkTarget__get_installname should append the SOVERSION, because the
installation set only contains the versioned libraries, not the
unversioned symlinks.
|
|
|
|
|
|
|
|
This reverts commit 1d90f180cd263d7f99039bada57b3655d6f399df.
|
|
|
|
|
|
This touch configure.in with a trivial space removal to force a
./bootstrap on the tinderboxes
This make ./bootstrap build a 'build-side' executable concat-deps
and modify LinkTarget.mk to use it instead of concat-deps.pl
The impact is a conservative x10 times improvement in both elapsed and
cpu time in overhead due to DEP generations.
time make -sr build in tail_build after removing
workdir/$INPATH/Dep/LinkTarget/Library/*
(all on Intel X3360, 2.83GHz, 4-core)
before:
elapsed: 116.5 cpu: 115.9
after
elapsed: 24.3 cpu: 23.9
base (no dep processing, i.e no rm of Dep/...)
elapsed: 17.5 cpu: 17.4
overhead ratio:
elapsed: 14.5x cpu: 15x
same with -j6
before:
elapsed: 47 cpu: 116
after:
elapsed: 20.8 cpu: 24
base:
elapsed: 17.5 cpu: 17.5
overhead ratio:
elapsed: 9x cpu: 15x
Note: for now the executable is generated directly in the source tree
(in solevn/bin ). That is not ideal. It and other similar polution
should be moved, to workdir most likely..
For now we just hide the mess under the .gitignore carpet
|
|
|
|
Reverse the dependency: now the component target depends (order-only) on the
library, which should ensure that gb_CppunitTest_add_component can set up
dependencies such that the unit test does not run concurrently with
a library loaded via UNO being overwritten.
Unfortunately this requires introducing another phony dummy library
target, on which the module can depend (because we don't want to register
the component target as a module target).
|
|
|
|
|
|
|
|
configurable at build time via ABOUTBOXPRODUCTVERSIONSUFFIX in
instsetoo_native/util/openoffice.lst
configurable at runtime via registry/data/org/openoffice/Setup.xcu
and ooSetupVersionAboutBoxSuffix property
|
|
|
|
|
|
so it does nothing on platforms where soversions are not used
|
|
|
|
|
|
This allows for setting a SONAME and version script, which is required
for UNO runtime libraries where we want to maintain a stable ABI.
The old build system did apparently not set an SONAME on MacOSX or
Solaris, so don't do that here either.
|
|
The option -P seems to be supported by a lot of ancient cp
implementations, let's hope it actually works.
|
|
It turns out even quite recent gdb (7.1) reportedly does not
support DWARF4, moreover while the produced binaries are smaller,
faster and make gdb initially use less memory, this quickly ceases
to the case as soon as one actually does anything in gdb.
Presumably the DWARF4 handling in current gdb is very inefficient
(http://sourceware.org/bugzilla/show_bug.cgi?id=13498).
This reverts commit 5214f7843ca48f2207945fbfe32ba164afe7b3aa.
Conflicts:
solenv/gbuild/platform/WNT_INTEL_GCC.mk
solenv/gbuild/platform/unxgcc.mk
|
|
|
|
This is a cherry-pick of Matúš's e2f30c078fcf26d481c2e90398b450f6c475a483
from the feature/gbuild branch, with the following modifications by
Stephan Bergmann <sbergman@redhat.com>:
* Adapt salhelper/Makefile to what all those Makefiles currently need to
look like.
* Do not remove salhelper/source/gcc3.map, instead add directly into it
what otherwise solenv/bin/addsym.awk would add to it on Linux.
* In salhelper/Library_salhelper.mk, add code that on Linux takes care of
the soname and symbol versioning required for backwards compatibility.
Solaris would need those features too, and its backwards compatibility
is thus currently broken. Also add a bad hack to create the soname
symlink (xxx.3 -> xxx) in the solver needed on non-Windows platforms (it
is a bad hack for now in that it e.g. is not removed by "make clean").
* In solenv/gbuild/platform/macosx.mk, add an even worse hack to set the
correct install name for libuno_salhelpergcc3.dylib.3, with a trailing
".3".
|
|
|
|
phase
|
|
|
|
|
|
gb_SourceEnvAndRecurse_STAGE is unset in CustomTarget's make process
when the parent make has been started in sourced environment. This leads
to all sorts of "interesting" failures because the gbuild classes and
platform stuff are not loaded...
|
|
|
|
|
|
|
|
|
|
|
|
This should fix the Windows build, while keeping the
gb_Helper_abbreviate_dirs_native working in custom targets.
|
|
This should first build, then run smoketest, then run subsequenttests.
|
|
So drop the GVER thing and -DVCL and -DNT351.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Formerly PRODUCTVERSION was used but that did not contain
the microversion digit. It is better to have it in MSI
ProductVersion, because a higher build number of a lower
microversion looked newer from the MSI point of view, and
update failed.
MSI ProductVersion is now major.minor.micro.build, e.g.
3.6.1.1206 (build number is taken from minor.mk)
|
|
+ added ProductName (a required parameter), It is LibreOffice
+ CompanyName is configurable, it is $(OOO_VENDOR), that is set
--with-vendor configure switch. If not set, then CompanyName is
empty string which is fine, too.
+ Binary file version and FileVersion string show the same value,
and version number is sane: main version + subversion + microversion
+ build number, such as 3.5.0.0
+ LegalCopyright text is the same as in the About box, year range
can be set in minor.mk
- removed archaic and overcomplicated version number calculating macros
- removed version.lst, copied relevant content to minor.mk
- removed binary RCDATA resources, I don't think they were of any use
|
