`buildcache create --rel`: deprecate this because there is no point in
making things relative before tarballing; on install you need to expand
`$ORIGIN` / `@loader_path` / relative symlinks anyways because some
dependencies may actually be in an upstream, or have different
projections.
`buildcache install --allow-root`: this flag was propagated through a
lot of functions but was ultimately unused.
* py-amici, py-python-libsbml: new packages
* Apply suggestions from code review
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* Swig and cmake are build-only dependencies
* cmake as a run dependency after all
* py-amici: default boost and hdf5 variants to True
---------
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
When building perl with posix jobserver, it seems to eat jobs, which
reduces parallelism to 1 in many cases, and is rather annoying. This is
solved in GNU Make 4.4 (fifo is more stable than file descriptors), but
that version is typically not available.
So, fix this issue by simply unsetting MAKEFLAGS for the duration of
./Configure. That's enough, and the build phase runs perfectly in
parallel again.
This switches the default Make build type to `build_type=Release`.
This offers:
- higher optimization level, including loop vectorization on older GCC
- adds NDEBUG define, which disables assertions, which could cause speedups if assertions are in loops etc
- no `-g` means smaller install size
Downsides are:
- worse backtraces (though this does NOT strip symbols)
- perf reports may be useless
- no function arguments / local variables in debugger (could be of course)
- no file path / line numbers in debugger
The downsides can be mitigated by overriding to `build_type=RelWithDebInfo` in `packages.yaml`,
if needed. The upside is that builds will be MUCH smaller (and faster) with this change.
---------
Co-authored-by: Gregory Becker <becker33@llnl.gov>
* Vendor ruamel.yaml v0.17.21
* Add unit test for whitespace regression
* Add an abstraction layer in Spack to wrap ruamel.yaml
All YAML operations are routed through spack.util.spack_yaml
The custom classes have been adapted to the new ruamel.yaml
class hierarchy.
Fixed line annotation issue in "spack config blame"
This ensures that:
a) no externals are added to the tarball metadata file
b) no externals are added to the prefix to prefix map on install, also
for old tarballs that did include externals
c) ensure that the prefix -> prefix map is always string to string, and
doesn't contain None in case for some reason a hash is missing
* libiconv can be provided by libc, so update packages which depend on
libiconv to require the iconv virtual instead
* Many packages need special consideration when locating iconv depending
on whether it is provided by libc (no prefix provided) or the libiconv
package (in that case we want to provide a prefix)
* It was also noticed that when an iconv external was provided, that
there was interference with linking (this should generally be handled
by Spack's compiler wrappers and bears further investigation)
* Like iconv, libintl can be provided by libc or another package, namely
gettext. It is not converted to a provider like libiconv because it
provides additional routines. The logic is similar to that of iconv
but instead of checking the provider, we check whether the gettext
installation includes libintl.
* Provide openmp from rocm-open-extras for roblas test
* Addressing the prechecks/audit/package-audits check
* Correcting style check errors.
* rocm-openmp-extras path veriable restricting for test
* Correcting the env variable to run_tests
* Guard use of OpenMP to make it optional in rocblas test
* Removing unused patch
* Patch to handle file reorg changes for the tests
* Correcting patch file name
* Limiting hipify-clang path to 5.4 and later
* Set hipify-clang path env in CMake
