Prepare for 2.13.0

[evaleev]

last preparations for 2.13.0 release that introduces @loriab's CMake harness. This will scrap the legacy harness(es).

[evaleev]

@loriab would appreciate a quick scan, as this aggregates your titanic efforts + option namespacing may have impact on psi4

[Copilot]

Pull request overview

This PR prepares for the 2.13.0 release by introducing a CMake harness to replace the legacy Autotools/Makefile build system. The changes involve systematic renaming of preprocessor macros to use consistent LIBINT_ prefixing, deletion of legacy build files, and updates to CMake configuration.

Changes:

• Renamed preprocessor macros throughout codebase: INCLUDE_* → LIBINT_INCLUDE_*, MULTIPOLE_MAX_ORDER → LIBINT_MULTIPOLE_MAX_ORDER, and other similar changes for consistency
• Removed legacy Makefile-based build system files (Makefiles, autoconf files, shell scripts)
• Updated CMake configuration to use LIBINT2_ prefix for user-facing options while maintaining LIBINT_ prefix for internal variables
• Updated documentation (INSTALL.md) to reflect new CMake option names

Reviewed changes

Copilot reviewed 52 out of 66 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
tests/unit/test-2body.cc	Updated macro from INCLUDE_ERI to LIBINT_INCLUDE_ERI
tests/unit/test-1body.cc	Updated macro from INCLUDE_ONEBODY to LIBINT_INCLUDE_ONEBODY
tests/hartree-fock/hartree-fock++.cc	Updated MULTIPOLE_MAX_ORDER to LIBINT_MULTIPOLE_MAX_ORDER
tests/eri/test.cc	Comprehensive macro renaming for all ERI types and derivatives
src/bin/libint/build_libint.cc	Updated all internal macro references to new naming scheme
src/bin/libint/buildtest.h	Added missing header includes
include/libint2/engine.impl.h	Updated preprocessor checks to use new macro names
include/libint2/engine.h	Updated documentation and macro references
include/libint2/deriv_map.h	Updated macro conditionals and added formatting improvements
include/libint2/cxxapi.h	Updated API availability checks
include/libint2/config.h.cmake.in	Comprehensive macro renaming in configuration template
export/fortran/*	Updated Fortran interface files with new macro names
cmake/modules/int_am.cmake	Updated CMake option handling with new naming convention
cmake/modules/int_orderings.cmake	Fixed variable naming bugs
cmake/modules/int_userreal.cmake	Updated variable names
src/lib/libint/populate.cmake	Fixed typo in comment
src/lib/libint/CMakeLists.txt	Updated internal variable references
CMakeLists.txt	Added LIBINT2_/LIBINT_ variable conversion logic
INSTALL.md	Updated all documentation to reflect new option names
.github/workflows/cmake.yml	Updated CI workflow with new option names
All Makefile files	Deleted legacy build system files
All autoconf files	Deleted legacy configuration files

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[loriab]

Thanks for all this! I realize now how out of practice I am with pure-cmake builds, so checking the migration guide as I go. I've generated a toy export and am testing that over at conda-forge. Now generating one that psi4 can consume ...

[evaleev]

I've generated a toy export and am testing that over at conda-forge. Now generating one that psi4 can consume ...

Here's what I found in adapting MPQC to 2.13.0:

• library target names have changed
• some PP tokens have changed, e.g. MULTIPOLE_MAX_ORDER -> LIBINT_MULTIPOLE_MAX_ORDER, INCLUDE_* -> LIBINT_INCLUDE_*, etc.

Nothing major otherwise. Of course the CMake option names now all start with LIBINT2_ (but some option renames were not just a concat).

Thanks again, I would have never done this. Neither could any agents do this. My attempt to use Claude to do cmake option renaming revealed that it is completely hopeless at groking nontrivial cmake code in int_am.cmake (straight text replace is fine, but programmatically-hidden instances stayed hidden)

[evaleev]

@loriab this seems to be in shape now to be merged (at least I managed to consume library from MPQC both via FC and find_package). BTW, re: DynamicVersion ... it seems that now I have to make a tag to change the version? Then make a release?

[loriab]

Are G12 ints working for you? I'm still investigating the case of the config below not happy compiling build_libint.cc

    -D LIBINT2_ENABLE_ERI=2 \
    -D LIBINT2_ENABLE_ERI3=2 \
    -D LIBINT2_ENABLE_ERI2=2 \
    -D LIBINT2_ENABLE_ONEBODY=2 \
    -D LIBINT2_ENABLE_G12=1 \
    -D LIBINT2_DISABLE_ONEBODY_PROPERTY_DERIVS=ON \
    -D LIBINT2_MULTIPOLE_MAX_ORDER=10 \
    -D LIBINT2_G12_MAX_AM=4 \
    -D LIBINT2_ERI_MAX_AM="5;5;3" \
    -D LIBINT2_ERI3_MAX_AM="6;6;4" \
    -D LIBINT2_ERI2_MAX_AM="6;6;4" \
    -D LIBINT2_MAX_AM="5;5;3"

                 from /psi/gits/libint2-upstream-wip/src/bin/libint/build_libint.cc:45:
/psi/gits/libint2-upstream-wip/build102/include/libint2/config.h:48: note: this is the location of the previous definition
   48 | #define LIBINT_MAX_AM 5
      | 
/psi/gits/libint2-upstream-wip/src/bin/libint/build_libint.cc:722: warning: "LIBINT_OPT_AM" redefined
  722 | #define LIBINT_OPT_AM G12_OPT_AM
      | 
/psi/gits/libint2-upstream-wip/build102/include/libint2/config.h:54: note: this is the location of the previous definition
   54 | #define LIBINT_OPT_AM 3
      | 
/psi/gits/libint2-upstream-wip/src/bin/libint/build_libint.cc: In function 'void try_main(int, char**)':
/psi/gits/libint2-upstream-wip/src/bin/libint/build_libint.cc:724:30: error: 'G12_MAX_AM' was not declared in this scope
  724 |   cparams->max_am("r12kg12", G12_MAX_AM);
      |                              ^~~~~~~~~~
/psi/gits/libint2-upstream-wip/src/bin/libint/build_libint.cc:725:34: error: 'G12_OPT_AM' was not declared in this scope
  725 |   cparams->max_am_opt("r12kg12", G12_OPT_AM);
      |                                  ^~~~~~~~~~
ninja: build stopped: subcommand failed.

[evaleev]

Are G12 ints working for you? I'm still investigating the case of the config below not happy compiling build_libint.cc

sheesh ... so many missed namespacing changes. pushed. G12 seems to generate/compile. G12DKH seems broken, but that code is experimental (so is G12 at this point ... no modern code uses it, keep them around since F12 is better formulated in terms of them).

[evaleev]

@loriab ok, now pushed

[loriab]

@loriab ok, now pushed

Thanks, I'll try that next. I started a PR #390 with one very minor fix I hit so far.

[loriab]

lgtm! I've a few optional changes at #391