Tarball for release 2.4.3 does not include all commits of the corresponding tag

[jpoiret]

The tarball leaves out commit f1f9ee78971b0e560a762a5e78d0dc8bc04fa682, which breaks compilation for macaulay2 with GCC 10.

[jpoiret]

After looking at this more closely, it seems that tag v2.4.3 was one year after tag 2.4.3, and that the release v2.4.3 on github links to a tarball of 2.4.3.

[ClementPernet]

This is very bad indeed.

• tag 2.4.3 was created soon after the release of 2.4.2, probably to declare the next release version. This should not have happened.
• Indeed upon releasing 2.4.2, the tag v2.4.3 was coined, but the tarballs were generated from the commit of 2.4.3. instead. As a consequence, there is a discrepancy between the v2.4.3 and the actual tar ball release which was used packaging in various systems. I can think of only one easy fix which will accommodate with the largely distributed tar balls erroneously based on 2.4.3 and not v2.4.3:
• remove this release entry in github, and create one for fflas-ffpack-2.4.3 at tag 2.4.3 with the same tarballs.
• add a tag v2.4.4 at v2.4.3 and release it as fflas-ffpack-2.4.2
• hopefully this mess will become less problematic, once we release 2.5.0, which we are working on now.

As for f1f9ee7, breaking compilation of macaulay2 with gcc10, I guess it deserves its own issue.

All apologies, as I am very likely the main culprit for this mess.

[ClementPernet]

Ok, I just updated the 2.4.3 release to point to the tag 2.4.3 and no longer v2.4.3. After a second thought, I will not create a 2.4.4 at the position of v2.4.3 to avoid unnecessary complications, since we will release 2.5.0, very soon.

[tornaria]

@ClementPernet: Note also that the tar.bz2 are named fflas_ffpack instead of fflas-ffpack. If you will keep doing bz2, maybe you can fix that to be fflas-ffpack as well.

[ClementPernet]

Well, no, this bz2 archive is produced for integration into sagemath, which requires this naming convention.