This reapplies 66f7540, which adds supports for hardlinks/junctions on
Windows systems where developer mode is not enabled.
The commit was reverted on account of multiple issues:
* Checks added to prevent dangling symlinks were interfering with
existing CI builds on Linux (i.e. builds that otherwise succeed were
failing for creating dangling symlinks).
* The logic also updated symlinking to perform redirection of relative
paths, which lead to malformed symlinks.
This commit fixes these issues.
There are other ways to enforce cray-pmi being loaded in environments
that use cray-mpich. This avoids breaking environments where this was
already the case and avoids forcing them to declare an external.
* AOCC and AOCL spack recipes for 4.1 release
* Fix broken checksum
* remove blank line
* Add missing `@when` for 4.1 only function
---------
Co-authored-by: vijay kallesh <Vijay-teekinavar.Kallesh@amd.com>
* Create a smartsim package
* rm ss 0.4.2
* no py upper bound, add build dep
* add setup_build_env
* add comment to find ml deps lower bounds
* Apply suggestions from code review
Correct dep versions, use `python_purelib`
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* remove the cuda/rocm vars
* point editors to bin deps version constraints
* Apply suggestions from code review
Loosen `py-smartredis` constraint, enforce `setup.cfg` py version
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* style
* rm rai lower bound
* lower bound setuptools
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
---------
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* add magma variant to strumpack
* clarify conflicts to be excuive to rocm/cuda, without it +magma+cuda fails as it is ~rocm
modified: var/spack/repos/builtin/packages/strumpack/package.py
* add missing depends_on for magma variant
#35042 introduced lazy hash parsing, but didn't remove a
few attributes from the parser that were needed only for
concrete specs
This commit removes them, since they are effectively
dead code.
Clone is a hard dependency as of HTTP-Message v6.44. This causes
problems in packages like Roary which depend on perl-http-message:
$ spack load roary
$ roary
Use of uninitialized value in require at /var/scratch/spack/opt/spack/linux-centos8-x86_64_v3/gcc-8.5.0/perl-http-message-6.44-lzp5th4jddd3gojkjfli4hljgem2nl26/lib/perl5/HTTP/Headers.pm line 8.
Can't locate Clone.pm in @INC (you may need to install the Clone module) (@INC contains: /home/aorth/lib ...
See: https://github.com/libwww-perl/HTTP-Message/blob/master/Changes
While spack does not yet provide binutils 2.41, they might still be
installed. However, building ffmpeg on x86_64 fails with multiple errors like
this:
./libavcodec/x86/mathops.h:125: Error: operand type mismatch for `shr'
also reported here: https://trac.ffmpeg.org/ticket/10405
Add amg2023 package
Consolidate existing amg and amg2013 packages (they reference the
same code) under the amg2013 name to minimize confusion between
amg2023 and amg2013.
Co-authored-by: Riyaz Haque <haque1@llnl.gov>
The heuristic for duplicate nodes contains a few typos, and
apparently slows down the solve for specs that have a lot of
sub-optimal choices to be taken.
This is likely because with a lot of sub-optimal choices, the
low priority, flawed heuristic is being used by clingo.
Here I split the heuristic, so complex rules that matter only
if we allow multiple nodes from the same package are used
only in that case.
* use full cuda_arch list
* update compiler requirement
* update boost requirements
* propagate +kokkos to legion in non-GPU cases
* add missing graphviz dependency for +doc
