- More robustly handle compiler version switching between QT4 and 5, and
mac/linux, and gcc/intel/clang
- Remove assumption about intel linker being in path
* Convert libmng to use CMake rather than autoconf
The autoconf script failed to recognize the intel compiler; it was
harwired to use gcc.
* Simplify cmake logic and remove unused variant
Add an optional 'submodules_delete' field to Git versions in Spack
packages that allows them to remove specific submodules.
For example: the nervanagpu submodule has become unavailable for the
PyTorch project (see issue 19457 at
https://github.com/pytorch/pytorch/issues/). Removing this submodule
allows 0.4.1 to build.
* Octave: moved the short description in its own paragraph
* Octave: patch mkoctfile.in.cc to avoid using compiler wrappers
* Added a check to ensure mkoctfile works correctly
* update libarchive and fix version of libarchive cmake dependency
* (at least) libarchive 3.3.3 dependency from cmake 3.15.0
* cmake depends on libarchive 3.1.0 if not specified differently
currently it is applied to cmake <3.15.0
Co-Authored-By: Adam J. Stewart <ajstewart426@gmail.com>
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* draco: update versions
+ Added versions 7.3.0 and 7.4.0.
+ Change several variants to be default TRUE since most consumers need these
variants enabled (eospac, lapack, parmetis, superlu-dist). Change variant name
for `+superlu_dist` to use hyphen instead of underscore. This makes the
variant name consistent with the spackage name for `superlu-dist`.
+ Clean up `depends_on` instructions and avoid specifying `type` when possible.
+ Provide patch files that are necessary for some machines (mostly Cray
machines).
* Remove trailing whitespace.
* Revert variant name to use underscore.
* add maintainer information.
Previously, the install stage would compile in things that were
disabled during the build_ext phase. This would also result in the
build pulling in locally installed versions of libraries that were
disabled. The install process doesn't honor the same command-line
flags that build_ext does, but does call build_ext again. Avoid the
whole issue by just writing the options to setup.cfg
Also, add the Imagemagick dependency for tests.
* pumi: sim version check, meshes via submodule, ctest
* Apply suggestions from code review
Co-Authored-By: Adam J. Stewart <ajstewart426@gmail.com>
* pumi: update comment on master version string
* pumi: description of simmodsuite_version_check variant
* pumi: add white space to variant description
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* git: add version 2.25.0 and fixup pcre dependency
pcre2 became optional in 2.14 and the default in 2.18. I noticed this
as git was compiling against the system pcre2 (spack was
specifying pcre as the dependency).
* missed a chunk from my internal repo
* Set netcdf-fortran to build serially with Intel compiler
This PR turns off parallel builds when the Intel compiler is used.
Builds with the Intel compiler will fail otherwise.
* Change how parallel build is handled
Use patch from netcdf-fortran project to turn off parallel buildi for
version 4.5.2.
* diffutils: Changed the handling of undeclared functions from warning to error.
* diffutils: Change the handling of warnings or error
* Delete '-Werror=implicit-function-declaration'
* Add '-Qunused-arguments'
Replace the deprecated ADIOS1 backend default with ADIOS2 default.
Disable sz since we do not need it and it conflicts with supported
version ranges between ADIOS2 and ADIOS1 if someone enables both.
* intel-tbb: Fix install names on Darwin
Intel-TBB's libraries on Darwin are installed with "@rpath" prefixed
to their install names. This was found to cause issues building the root
package on Darwin due to libtbb not being found when running some of the
generated tools linking to it.
Follow example from other packages with the same issue and fixup up install
names for intel-tbb post install.
* intel-tbb: fix flake8 errors
* Dirty hack to fix#14148
* A better way of checking if a package is taken from system
* Update var/spack/repos/builtin/packages/qt/package.py
Co-Authored-By: Adam J. Stewart <ajstewart426@gmail.com>
* Update qt/package.py
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* Add the py-merlinwf package
* Fix importlib-resources package name for spack naming convention.
* Add build to dependencies and add updated versions.
* Remove pytest-runner dependency.
* Fix typo.
* Add the py-tabulate dependency.
* Add sha256 for version 1.0.0
* Change to maestro version 1.1.5.
* Increase to version 1.0.4.
* Bump maestrowf version and prepare for new pypi version.
* Add sha256sum for version 1.1.5
* Add version 1.1.1.
Update maestrowf version to 1.1.7
* Add versions 1.0.5, 1.1.0, 1.1.1 and potential 1.2.0.
* Add version 1.2.0 and when on maestrowf@1.1.6.
* Add version 1.2.2 , remove 1.2.1 and 1.1.0.
* Update var/spack/repos/builtin/packages/py-merlinwf/package.py
Co-Authored-By: Adam J. Stewart <ajstewart426@gmail.com>
* Update var/spack/repos/builtin/packages/py-merlinwf/package.py
Co-Authored-By: Adam J. Stewart <ajstewart426@gmail.com>
* Remove mysql variant until new mysql interface module is enabled.
The mysql code may be removed.
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* Get py-torch to build caffe2
This PR gets the py-torch package to build with caffe2, and closes
issue #14576. If building on a machine with CUDA but no GPU the build
will try to build with all compute capabilities. Older compute
capabilities are not supported so the build will fail. The list of
capabilities can be passed to the build using values set in the
cuda_arch variant. Likewise, conflicts are also set to catch if the
unsupported capabilities are listed in cuda_arch.
This PR also sets version constraints on using an external mkldnn for
newer versions. Currenly, only versions up to 0.4 use an external mkldnn
library. Also, the cuda variant is set to True, which restores
previous behavior.
* Update var/spack/repos/builtin/packages/py-torch/package.py
Fix typo.
Co-Authored-By: Adam J. Stewart <ajstewart426@gmail.com>
* Adjust conflicts
This commit adjusts the conflicts. There is an issue with the
cuda_arch=20 conflicts directive as there is a conflicting dependency
with any version >=1.1 and a cuda_arch=20 dependency specified in
CudaPackage that gets trapped first.
* Use a common message for conflicts
This commit adds a variable to contain the bulk of the message stringi
for the cuda_arch conflicts. This is used along with a version string
in the conflicts directives messages.
* Fix the strings
- Use a multiline string for the cuda_arch_conflict variable.
- No need for format() in the msg value.
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
Openblas target is now determined automatically upon inspection of
`TargetList.txt`. If the spack target is a generic architecture family
(like x86_64 or aarch64) the DYNAMIC_ARCH setting is used
instead of targeting a specific microarchitecture.
* Try to switch to a newer fork of ftgl
* Allow ROOT to be more flexible about ftgl versions
* Turn ftgl into a CMakePackage
* Update ROOT ftgl dep since 2.1.3 isn't a thing anymore
* Please flake8
* Try to bring back the doc variant
* Comment it out instead of removing it
* Fix root+x breakage from #11129
* Separate out +opengl breakage
* Not strictly X11-related, but more breakage from #11129
* Another X11 breakage found while building 6.08.x
* Don't put system headers in SPACK_INCLUDE_DIRS + deduplicate
* xextproto is only a dependency in +x builds
