* Added a package for the MDAnalysis toolkit.
* Added Ninja as the generator for LBANN, Hydrogen, and Aluminum. Also
fixed a bug in how the Conduit variant is included in LBANN. Added
new versions for Hydrogen and LBANN.
* Updated lbann package so that the CMAKE parameters are set when the
variant is on and off.
Fix the install of the single header amalgate in catch2.
The public API for catch < 2.3.0 was "#include <catch.hpp>".
For 2.3.0+ it is "#include <catch2/catch.hpp>"
* Spades: Add version 3.13.0
In version 3.13.0 the CmakeLists.txt dir is at a different location than
in other versions. Versions up to 3.12.0 are located in "src" whereas
3.13.0 is a level deeper at "assemblers/src".
* Remove git and add new version hash
- all multimethod tests are now run for both `multimethod` and
`multimethod-inheritor`
- do this with a parameterized fixture (pkg_name) that runs the same
tests on both
- Since early Spack versions, the SpecParser has (weirdly) been
responsible for initializing Spec fields.
- This refactors initialization to take place in Spec.__init__, as it
probably should have originally.
- This makes the code easier to read, the parser easier to understand,
and removes the use of __new__ in the parser to initialize the Spec.
- This also makes it possible to make a completely empty Spec with
`Spec()` -- this is an abstract Spec that will match anything.
* Add mising QE releases.
* Update QE dependencies and conflicts.
* QE 6.3 install method requires a patch to work properly.
* include QE git develop branch
* QE HDF5 needs Fortran support. HDF5 versions prior to 1.8.16 leads to QE runtime errors.
* Number of MKL related conflicts for QE.
* Flake8 fixes for QE for hdf5 lines.
* Pass in all compilers properly to QE configure
* Fixed external BLAS/LAPACK linkage in QE.
* Library search was problematic in QE.
* Add section headings to QE package.
* External FFTW compatible libraries properly detected in QE.
* Flake8 fixes for QE
* QE configure defaults to False for ELPA, thus Spack package should as well.
* QE depends on FFTW3 provider instead of FFTW explictly.
* Primary QE repo is gitlab.com, not github.com. Github mirror lags behind gitlab site.
* Support both 6.3-backports branch and develop branch in QE.
* QE patch needed for proper MKL detection. MKL_ROOT is detected and preferred, still falls back to finding first MKL installation in /opt/intel.
* Use FFTW-API provider versioned interface for specify QE FFT dependency.
* QE FFTW conflicts no longer need thanks to FFTW-API provider versioned interface.
* Flake8 3.6.0 fix in support of https://github.com/spack/spack/pull/9853
* Fix QE conflict with ilp64 variant in support of https://github.com/spack/spack/pull/9853
* Remove custom url_for_version for QE package and use builtin url instead. In support of https://github.com/spack/spack/pull/9853
* Use FFTW builtin flags for QE instead of custom Python code. In support of https://github.com/spack/spack/pull/9853
* Fix typo in QE package comments. In support of https://github.com/spack/spack/pull/9853
* Future proof QE package patches in anticipation of QE >= 6.3.1. In support of https://github.com/spack/spack/pull/9853
* Adding version 2.0.12 of numactl.
* Adding new versions of hwloc.
* Adding NVML variant to hwloc.
* Adding gl variant to hwloc.
* Updating hwloc numactl dependency and gl dependency.
* Using gl virtual package in hwloc.
Clang 8.0 now supports openmp 4.5 and many commerical compilers also support
openmp >= 4.0. This commit removes the openmp version guard and rely on CMake
to throw errors out when openmp version requirement is not satisified.
It also add options for precision control.
* CGNS: Add option for 64-bit integers
Added the `int64` variant which will build the library using 64-bit integers for certain values. This gives the capability to have models with more than 2 billion cells and/or nodes.
Beginning with CGNS-3.1.0, two new typedef variables have been introduced to support 64-bit mode. The `cglong_t` typedef is always a 64-bit integer, and `cgsize_t` will be either a 32-bit or 64-bit integer depending on how the library was built. Many of the C functions in the MLL have been changed to to use `cgsize_t` instead of `int` in the arguments. These functions include any that may exceed the 2Gb limit of an` int`, e.g. zone dimensions, element data, boundary conditions, and connectivity. In Fortran, all integer data is taken to be `integer*4` for 32-bit and `integer*8` for 64-bit builds.
* EXODUS: Update version and dependencies
Allow pulling master version from git. Note that I hope to have better versioning "soon"
Add requirement to use Netcdf-4.6.1 or later as those versions no longer require changing the `maxdims` or `maxvars` defines in the netcdf.h include file. With these versions, Exodus can use an unmodified NetCDF. These versions of NetCDF are also better than previous versions (code quality, performance, robustness).
It is still possible to run exodusII with previous NetCDF versions, but I don't know how to specify that the dependency where maxdims/maxvars variants are only needed for versions prior to 4.6.0.
I also removed the explicit dependency on HDF5 as that dependency should be applied through the NetCDF dependency.
* Remove white-space on blank line
* Provide a new spackage for py-pint.
Pint is a Python package to define, operate and manipulate physical quantities:
the product of a numerical value and a unit of measurement. It allows arithmetic
operations between them and conversions from and to different
units. https://pint.readthedocs.io/en/latest/
* Switch hash from md5 to sha256.
Add several new versions so that current functionality can be accessed. Current version is now 1.5.13.
Update `url`, `list_url`, and `list_depth` so spack can find available versions
* PNETCDF: Update with current versions
Added latest versions. (1.10.0, 1.9.0, 1.8.1)
* PNETCDF: Revert version change in url
Reverted the change in the url variable version since not required.
