v2024.06.0
Version v2024.06.0 -- Release date 2024-06-27
- Important Notes:
- External users of the code will need to supply config files for tpl-manager to find system libraries correctly. Steps to do this are detailed in the external user build guide.
Notable changes include:
-
New features / API changes:
- Added MFV hydro from Hopkins 2015 with extension for ALE options.
- Adding optional user specified smoothing scale method for SPH, FSISPH, and CRKSPH.
-
Build changes / improvements:
- PYBind11 libraries no longer depend on the structure of the PYB11 source directory.
- CMake interface for adding PYBind11 target libraries is modified to more closely match how C++ libraries are created.
- Multiple Spheral Python modules / CMake targets can be specified for a single directory.
- KernelIntegrator and FieldList directories are divided into 2 modules / targets.
- tpl-manager.py will no longer use generic x86_64 configs for non LC systems. Users will be required to supply their own configs for pointing spack at external packages.
- Spack version is increased from 0.19 to 0.22.
- Spack upstream is updated.
- Removed the python 3 module load for the Gitlab CI to fix an issue with pkg-config changing.
- Zlib target and TPL cmake file is removed.
- PYB11Generator repo is updated.
- Spack config and package files inside Spheral are updated to accommodate Spack 0.22.
- Package recipes for py-numpy-stl, py-pillow, py-pipreqs, td, and tk are removed.
- Versions for python dependencies in the Spheral spack recipe are fixed and updated (in some cases).
- PYBind11 libraries no longer depend on the structure of the PYB11 source directory.
-
Bug Fixes / improvements:
- Corrected an erroneous VERIFY in the P-alpha porosity constructor (with Fields of porosity and sound speed) that forced runs to stop even with correct input parameters
- Fixed a bug in the standard ASPH hydros (ASPH, SolidASPH, and RZ varieties) that gave incorrect results. FSI ad CRK models with ASPH smoothing scales were OK, but standard
SPH using ASPH smoothing scales were simply incorrect for non-unit aspect ratio H's. Also added ATS tests to help catch such errors going forward.