summarylogtreecommitdiffstats
path: root/PKGBUILD
AgeCommit message (Collapse)Author
2016-09-05Dirty fix for unittests failingLuchesar V. ILIEV
2016-06-30D18035 committed and not needed any moreLuchesar V. ILIEV
2016-06-29Sync .SRCINFOLuchesar V. ILIEV
2016-06-29Update D18035Luchesar V. ILIEV
2016-06-11Sync .SRCINFOLuchesar V. ILIEV
2016-05-23Sync .SRCINFOLuchesar V. ILIEV
2016-05-23Update D18035Luchesar V. ILIEV
2016-05-09Sync .SRCINFOLuchesar V. ILIEV
2016-05-09Add make checkLuchesar V. ILIEV
2016-05-09Update D18035Luchesar V. ILIEV
2016-03-16Sync .SRCINFOLuchesar V. ILIEV
2016-03-16Update D18035Luchesar V. ILIEV
2016-03-10Rebase D18035Luchesar V. ILIEV
2016-03-10Update the add-gcc-abi_tag patchLuchesar V. ILIEV
The Sema part of D12834 has already been incorporated upstream with r263015. Added here is the respective mangler part from D18035. References: https://llvm.org/bugs/show_bug.cgi?id=23529 http://reviews.llvm.org/D12834 http://reviews.llvm.org/D17567 http://reviews.llvm.org/D18035
2016-03-09Sync .SRCINFOLuchesar V. ILIEV
2016-03-09Rebase D12834Luchesar V. ILIEV
2016-02-27Sync .SRCINFOLuchesar V. ILIEV
2016-02-27Fix for r262066: lto.h in libs pkgLuchesar V. ILIEV
2016-02-16Sync .SRCINFOLuchesar V. ILIEV
2016-02-03Sync .SRCINFOLuchesar V. ILIEV
2016-02-03Rebase D12834Luchesar V. ILIEV
2016-02-03Move the OCaml bindings check back to build()Luchesar V. ILIEV
The most time consuming operation is preparing the working copies, and by checking in prepare() we still can't spare that time for the users. However, checking in prepare() may break unnecessarily `makepkg -o'. Finally, even just logic calls for this check to be performed in build(), where it also did originally belong. This partially reverts commit 11fbb41699576e0c38ee39742008dd34fa74607f.
2016-01-30Sync .SRCINFOLuchesar V. ILIEV
2016-01-30Fix pkgver() for CMakeLuchesar V. ILIEV
2016-01-20Update D12834 patch and rebaseLuchesar V. ILIEV
2016-01-14Push version to 3.9.0svnLuchesar V. ILIEV
2015-12-13Add support for GCC's attribute abi_tagLuchesar V. ILIEV
With the release of GCC 5.1, libstdc++ has started using the abi_tag attribute, documented here: https://gcc.gnu.org/onlinedocs/gcc/C_002b_002b-Attributes.html Arch Linux has switched to the new ABI on 2015-12-10: https://www.archlinux.org/news/c-abi-change/ This set of patches tries to adapt Clang to this new dual ABI model when linking to libraries that provide interfaces based on it: http://lists.llvm.org/pipermail/cfe-commits/Week-of-Mon-20151207/144636.html For further information, see: * https://llvm.org/bugs/show_bug.cgi?id=23529 * http://reviews.llvm.org/D12834 * https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=797917
2015-12-13Do `make ocaml_doc` after `make all`Luchesar V. ILIEV
docs/cmake_install.cmake used to fail if `make all` was run without running `make ocaml_doc` in advance. However, this doesn't quite make sense and was probably just a transient problem. On the other hand, running `make ocaml_doc` before `make all` seems to cause some random errors of different kind. For these reasons, reverse their order.
2015-12-13Update the incompatible OCaml bindings checkLuchesar V. ILIEV
It turns out that even same version bindings can cause problems, so assume any installed system-wide bindings as incompatible. Also, check earlier so users don't waste time unnecessarily.
2015-11-14Sync .SRCINFOLuchesar V. ILIEV
2015-11-14Correct the Python precompileLuchesar V. ILIEV
2015-11-14Sync .SRCINFOLuchesar V. ILIEV
2015-11-14Yet another update for the Clang Static AnalyzerLuchesar V. ILIEV
2015-11-13Sync .SRCINFOLuchesar V. ILIEV
2015-11-13Another fix for the Clang static analyzer...Luchesar V. ILIEV
...due to changes upstream. While here, try to make the code more readable and easy to manage.
2015-11-10Sync .SRCINFOLuchesar V. ILIEV
2015-11-10Sync .SRCINFOLuchesar V. ILIEV
2015-11-10Update for r252474 and r252489 from upstreamLuchesar V. ILIEV
2015-11-05Update for r251411 and r252093 from upstreamLuchesar V. ILIEV
2015-10-22Revert "[CMake] Fix add_llvm_external_project()"Luchesar V. ILIEV
The patch has been accepted upstream as r251001. This reverts commit 1ad73e4035d69f7549270029919550613c7e3b61.
2015-10-21Sync .SRCINFOLuchesar V. ILIEV
2015-10-21[CMake] Fix add_llvm_external_project()Luchesar V. ILIEV
r250835 unintentionally discarded the optional parameter to the add_llvm_external_project() macro that may point to a path when the said path is different from ${name}. This should fix it by passing ${ARGN} on to add_llvm_subdirectory(). The problem manifests itself with e.g. add_llvm_external_project(clang-tools-extra extra) from clang/tools/CMakeLists.txt References: http://reviews.llvm.org/D13783 http://llvm.org/viewvc/llvm-project?view=revision&revision=250835
2015-10-10Sync .SRCINFOLuchesar V. ILIEV
2015-10-10Add official package maintainer as contributorLuchesar V. ILIEV
Since we merge some changes from the official package, it is proper to give attribution. Also, sort the contributors alphabetically.
2015-10-10Use more elegant pkg-config commands for libffiLuchesar V. ILIEV
2015-10-10Remove the shlib patch for LLVM bug 24157Luchesar V. ILIEV
LLVM r249862 should finally fix the shared library not exporting all public symbols when built with CMake. Ref: https://llvm.org/bugs/show_bug.cgi?id=24157
2015-09-09Put all packages in a package groupLuchesar V. ILIEV
This should allow for easy installation of the complete LLVM toolchain. The only package that is not part of this group are the OCaml bindings.
2015-09-09Add the missing compiler-rt runtime librariesLuchesar V. ILIEV
While the libraries are part of the official Clang package, we opt for putting them in a separate package, which is how upstream does at http://llvm.org/apt/. This commit solves issue #5.
2015-09-06Update for the shared library changes upstreamLuchesar V. ILIEV
Due to a bug upstream, libLLVM wouldn't export all expected symbols on Linux. We've previously implemented some awk-wardish magic to fix this on our side, as explained in issue #2. However, two recent commits upstream, llvm-mirror/llvm@10add60 and llvm-mirror/llvm@f5148eb, have tried to fix the original bug, which, unsurprisingly, also broke our now unnecessary fixes. This commit basically removes those fixes, with the notable exception of one: ConvertUTF*, getNumBytesForUTF8, and isLegalUTF8* are still not being exported, so we must continue patching tools/llvm-shlib/CMakeLists.txt as before.
2015-09-04Put common code in reusable functionsLuchesar V. ILIEV
Some code is shared between the different package_* functions, so it's better to have it in one place for cleaner code and easier support.