"make check" fails for provided Libint 2.6.0 source tarballs

I tried running "make check" after configuring with ./configure --prefix /tmp/ --enable-fortran and building with make -j 9, but I ran into problems like:

../../include/libint2/./engine.impl.h:627:3: note: suggested alternative: LIBINT2_MAX_AM_default
../../include/libint2/./engine.impl.h:627:3: error: LIBINT2_MAX_AM_default1 was not declared in this scope
   BOOST_PP_LIST_FOR_EACH_PRODUCT(
   ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The same problem occurs with multiple tarballs (I tried libint-v2.6.0-cp2k-lmax-4.tgz and libint-v2.6.0-cp2k-lmax-7.tgz).

There's a fix for this issue that was merged via evaleev/libint#149, but this requiring re-doing the "make export"... Is there a way to run the tests with the source tarballs provided here?

If I apply the fix from evaleev/libint#149 to the original Libint 2.6.0 sources, configure as done in

libint-cp2k/Jenkinsfile

Line 55 in ef73545

def configure_libint(lmax) {

, run "make export" and then rerun the configure/make/make install cycle using the obtained source tarball, things seem to be working as expected (make check passes without problems).

Sorry, this issue was unnoticed. Since this bug was fixed in libint, there are no longer any problems with newer release of libint-cp2k, I think. I close this but feel free to reopen if this issue remains.

@pseewald Do you have a pointer to the fix in Libint?

Ah, nevermind, I pointed to the fix myself... 🤦 (evaleev/libint#149)

Thanks for following up!

Well, it's still a problem with the 2.6.0 libint-cp2k, it's a bit unfortunate that libint upstream does not seem to think this is worth issuing a 2.6.1 for. Could this be patched in the Jenkinsfile for libint-cp2k?