Some problems with library on Windows

[giordano]

There are currently two orthogonal issues with the library on Windows (spotted in jump-dev/Ipopt.jl#327 (comment)):

• the soversion of the library, fixed in src/Make.inc is stalling:

  libblastrampoline/src/Make.inc

  Lines 26 to 28 in b829b0e

  	LBT_SOVERSION_MAJOR := 5
  	LBT_SOVERSION_MINOR := 0
  	LBT_SOVERSION_PATCH := 2

  but the latest release at the moment is v5.1.1

• libblastrampoline/src/Makefile

  Lines 45 to 46 in b829b0e

  	$(builddir)/$(LIB_FULL_VERSION): $(MAIN_OBJS)
  	@$(call PRINT_CC,$(CC) -o "$@" $(call IMPLIB_FLAGS,$@) $(LBT_CFLAGS) $(SONAME_FLAG) $^ -shared $(LBT_LDFLAGS))

  generates the import library for $@, which is LIB_FULL_VERSION, which is a bad idea because it means the ABI is broken in every single version (well, every single version where the soversion is actually bumped, see point above)

An overall solution is probably to stop generating the library with the full name on Windows, or at least generating the import library for LIB_MAJOR_VERSION instead of LIB_FULL_VERSION.

It remains the problem of actually remembering to update the soversion in src/Make.inc when doing a new release.

CC: @ViralBShah

[ViralBShah]

So should we sync the version, or just have only the major version? I assume we need Julia 1.8.1 with this fix, right?

[giordano]

So should we sync the version, or just have only the major version?

In any case we should generate the import library for LIB_MAJOR_VERSION, not for LIB_FULL_VERSION, that's a really bad idea. There is the question of whether to keep LIB_FULL_VERSION around.

Going forward it'd be great the have the soversion actually match the "version number" of the package, otherwise it's an inconsistent mess.

I assume we need Julia 1.8.1 with this fix, right?

Yes, but fixing this issue will break the ABI of the library on Windows.

[ViralBShah]

Will this fix the null pointer issue in mumps_seq and ipopt_jll on windows? Is it still something else?

[giordano]

I don't think this is the culprit there, but it is a compatibility problem

[giordano]

Will this fix the null pointer issue in mumps_seq and ipopt_jll on windows? Is it still something else?

For the record, that issue is tracked by #89, and may be related to the confusion between the libraries available on Windows.