The source is managed as a set of files in an FTP URL. Since there
are no FetchStrategy's implemented for this setup, the files are
managed as a set of Resource's
Add dependencies to CMake and GnuTLS which would otherwise be
picked up from the system
* Add gnutls as explicit dependency of CMake
* Add new package libidn2 as explicit dependency for gnutls
- `svn info` prints different results depending on the system locale
- in particular, Japanese output doesn't contain "Revision:"
- Change Spack code to use XML output instead of using the human output
* LBANN: add python dependency and required python library dependencies
* LBANN: add version 0.99 as alias for develop branch
* LBANN: build Conduit library by default
* LBANN: CUB is included with Hydrogen for later versions
* New package: py-graphviz (dependency for LBANN)
* New package: py-texttable (dependency for LBANN)
* New package: py-onnx (dependency for LBANN)
* New package: py-typing-extensions (dependency for py-onnx)
* Conduit: add version 0.4.0
* NCCL: Add versions 2.4.6-1 and 2.4.2-1
* Protobuf: Add versions 3.7.1 and 3.7.0
* py-protobuf: add version 3.7.1
* py-typing: add version 3.6.4
* cuDNN: add versions. The source differs depending on the
architecture, so the versions include a suffix to encode this.
For example this adds 7.5.1-10.1-x86_64 as well as
7.5.1-10.1-ppc64le
* cuDNN: Newer versions require newer versions of CUDA
* Hydrogen: add configure options for CUB and Aluminum
The --enable-relax-coord-bound option was added in 1.8.0. NetCDF
4.6.3 enabled relax coords by default and required it from PNetCDF.
PNetCDF 1.10.0 enabled it by default but for versions between 1.8.0
and 1.10.0 it should be enabled explicitly.
* megadock: new package at 4.0.3
* megadock: remove CudaPackage redundancies
* megadock: move env vars to build targets, use cuda_arch variant
* megadock: flake8 fixes, more flexible mathlib
* Fix py-yt for Python 2 by requesting older py-ipython version.
* Add conflicts statement to py-ipython because version 7: is not
compatible with python@2.7
f242f5f removed spec.architecture.platform_os, which was used to test
an option in the 'm4' package. The attribute is now accessed as
'spec.architecture.os'.
* Add versions 3.27.{0,1,2} and 3.28.0
* Add url_for_version: the URL format includes the year so this
function maps version ranges to years
* Restore patches for versions removed in 84c547c: this commit
removed versions before 3.26.0 due to a security issue with FTS
support. Add a +fts variant (default true) that enables FTS
support along with a conflict for +fts with earlier versions.
- Codecov cannot handle as many coverage reports as we are generating
- as a result, our PR coverage pages have been broken for a while, and
it's hard to tell people where to enhance their testing in PR reviews.
- Scale back to only running coverage for 3.7 and 2.7 unit tests
- This is *probably* better. We run the build tests for good measure,
but we do not need to evaluate them for coverage. The coverage reports
are about unit tests.
Add fixes to support multiple installs and dependents using a subset
of IntelPackage functionality.
* Update IntelPackage to only return scalapack libraries if the root
spec depends on MPI: scalapack requires MPI to be mentioned as a
dependency in the DAG. Package builds using intel-mkl for its
blas/lapack implementations but not for scalapack were failing to
build.
Ideally it would be possible to ask if any of the packages in the
DAG are actually requesting the scalapack functionality provided by
the IntelPackage and only return scalapack libs in that case, but
that is not easily done at this time.
Fixes#11314Fixes#11289
* set HOME when the intel silent installer is run. This prevents the
installer from using the ~/intel directory (which can cause
conflicts for multiple installs of the same IntelPackage)
Fixes#9713
Use new `module` function instead of `get_module_cmd`
Previously, Spack relied on either examining the bash `module()` function or using the `which` command to find the underlying executable for modules. More complicated module systems do not allow for the sort of simple analysis we were doing (see #6451).
Spack now uses the `module` function directly and copies environment changes from the resulting subprocess back into Spack. This should provide a future-proof implementation for changes to the logic underlying the module system on various HPC systems.
