Changelog for LHAPDF-devel-6.5.4-51.63.x86_64.rpm :

* Mon Apr 10 2023 Atri Bhattacharya - Update to version 6.5.4:
* Add __repr__ methods to Python wrappers.
* Add keys() and keys_local() methods to Info classes.
* Fix PDFUncertainty and PDFErrInfo object lifetimes in the Python interface.
* Add setname_nmem variants of methods for making PDFInfo and AlphaS objects.
* Introduce MemType and MemDesc flags, used by preference to PdfType and PdfDesc.
* Convert MPI interface to use MPI v3.
* Avoid silent compilation failures during the python extension build.
* Replace deprecated egrep with grep -E in lhapdf-config and other places.
* Change nmem type in mk function signatures from int to size_t, and fix several other signed/unsigned warnings.
* Add CXXFLAGS and LDFLAGS propagation to the Python-extension build script.
* Fix bug in in precomputed grid-differences that would lead to lookups outside array bounds.- Drop LHAPDF-python-platlib.patch: fixed upstream.
* Thu Sep 01 2022 Atri Bhattacharya - Update to version 6.5.3:
* Fix linker arguments in Python-extension build.- Drop LHAPDF-python-build-fix.patch: incorporated upstream.
* Wed Aug 24 2022 Atri Bhattacharya - Update to version 6.5.2:
* Fix range of member iteration when computing correlations with the next extended-errorset scheme
* Update internal yaml-cpp version to fix compiler warnings.
* Complete distutils removal on main.
* Fix bug in lepton PDF lookup.
* Fail with an actually helpful exception message if lhapdf.conf can\'t be found.- Add LHAPDF-python-platlib.patch -- Correct system platlib determination in configure.ac.- Add LHAPDF-python-build-fix.patch -- Fix python builds due to an unset env variable; patch taken from upstream git commit and re-based for current version.- Move /usr/bin/lhapdf binary to -devel pkg; this does not have anything to do with the LHAPDF python bindings but rather is a general utility script using standard python libs.- Disable timestamps in doxygen generated docs.- Run spec-cleaner for minor cleanups.
* Sun May 29 2022 Atri Bhattacharya - Update to version 6.5.1:
* Add uncertainty testing to release process.
* Fix typo in PDFSet.cc use of default uncertainty CL in percent.- Changes from version 6.5.0:
* Remove distutils dependency from configure, fall back to using setuptools if possible in Python extension build, and update Python version detection macros.
* Apply thread locality fixes to caches, active-set management in LHAGlue, and non-MPI file-IO objects.
* Add PDFSet::uncertainties() methods to compute multiple observable uncertainties at once.
* Chain the in-place and return by value PDFSet::uncertainty() methods in the more efficient order.
* Lazily cache the PDFSet error-info struct.
* Add a parser for the extended error-type string format, and start implementation of the extended uncertainty computation using it.
* Use the PDFSet::lhapdfID() function by preference to return a PDF::lhapdfID(), thereby avoiding the index-file lookup if possible.
* Assemble the PDF-set index from _all_ pdfsets.index files found in the search paths, with earlier appearances taking precedence: allows use of partial indices in overload directories.- Delete some generated example binaries from the doc pkg.- Run fdupes to link some documentation files (BuildRequires: fdupes).
* Sat Jan 15 2022 Atri Bhattacharya - Delete cython generated cpp file so it can be regenerated by the right flavour of Cython.- Add BuildRequires: python-Cython, needed for above.
* Thu Sep 30 2021 Atri Bhattacharya - Update to version 6.4.0:
* Restructure internal memory storage to increase performance and simplify caching common computations for different particles. Most notably replace KnotArray1F and KnotArrayNF with single class KnotArray.
* Precompute coefficients for polynomial interpolators at init time.
* Add specialised method to interpolators, to evaluate pdf\'s for all common particles (-6,...-6,21) and allow for the caching of common computations withoutthe need of a complicated cache.
* Update the fast ASCII NumParser to always use the C locale.
* Enable out-of-source builds, avoiding sensitivity to pre-installed versions during the build, and converting the Cython syntax to Py3-style. -atch.
* Bugfix to Python wrapper object parentage handling.
* Add a PIDs::PIDCode enum (and PIDs enum with equivalent signature in Python) for more semantic access to particle ID codes.
* Update embedded yaml-cpp code to version 0.6.3.
* PDFSet::uncertainties() now allows alternative=True for non-replica sets (in which case it does nothing).
* lhapdf script bugfixes and improvements, including only complaining about failed URLs after
*all
* attempts have failed.
* Add matching by ID code to the lhapdf list and show commands.
* Mon Feb 15 2021 Atri Bhattacharya - Fix building for multiple python flavours.- Copy doc files (built only for default python3 flavour) to buildroot during %install.
* Fri Jul 17 2020 Atri Bhattacharya - Update to version 6.3.0:
* Provide setter methods on PDF for the flavour list and setPositive behaviour.
* Convert the AlphaS raw pointer on PDF to a std::unique_ptr.
* Update the cache strategy for \'replays\', and make cache size and search/update strategy configurable at runtime.
* Extend caches to multi-level, searching through the N previous calls for x and Q2 independently.
* Make the caches fully thread-safe rather than just initialization-safe, by use of maps on thread IDs.
* Extend to cross-flavour, subgrid-specific x and Q2 caches.
* Use the x and Q2 knot hashes in cache-flushing tests.
* Add a hash function computation and comparator to the KnotArray1F.
* Convert caching struct acquisition to use a Meyers Singleton pattern, which guarantees thread-locality.
* Separate caching of x and Q2 variables in the LogBicubicInterpolator: evidence suggests that these may be updated at very different rates.
* Add simple caching to the LogBicubicInterpolator -- will cache contiguous calls of the same (x,Q).- Rebase sover.diff for update.- Build API documentation and package them as part of a new doc package.- Add BuildRequires: python-setuptools to allow proper postprocessing of the python-LHAPDF package when building.- Drop obsolete conditionals and Group tags.
* Tue Apr 28 2020 Atri Bhattacharya - Use python-rpm-macros to allow building python mod for all supported versions of python; copy source into a different directory for each python version and configure, build, and install them correctly.- Move python dependent binary to python subpackage (python3 only).
* Tue Aug 20 2019 Atri Bhattacharya - Update to version 6.2.3:
* Add a couple of extra trim() calls to path strings given to the Fortran API.
* Apply lhapdf script improvement from Ivan Razumov / Genser.- Changes from version 6.2.2:
* Don\'t pop the final path entry off in the pathsPrepend and pathsAppend functions: the compile-time fallback may have been disabled.
* Add a \'show\' option to the lhapdf script.
* Adapt MPI file-reading patch from Stefan Hoeche. Enable with - -enable-mpi to configure.
* Rewrite the lhapdf data-manager script to use argparse, for compatibility with Python 3.
* Apply compiler optimisation bypass \"volatile\" hack to default data path insertion. Thanks to Zahari Kassabov.
* Add legend PDF name aliasing to plotpdfs
* Explicitly catch x > xMax error in ContinuationExtrapolator.
* Don\'t require the presence of Content-Length in HTTP header.
* Utils.h: Fix separator bug in join() function.
* Remove \'const\'s from return-by-value types in KnotArray.h, to keep icpc happy.
* Add lhapdf_lambda4/5 functions in the Fortran interface.
* Remove unwanted pid argument from the Fortran xfx
*_stdpartons functions.
* Move Config::get() static function from header to .cc, since a singleton needs to only \'live\' in one compilation unit.- Drop LHAPDF-pdfset-download-url-update.patch: incorporated upstream.- Replace env-based hashbang with one explicitly referring to the correct binary.