Why3 1.7.0

[silene]

No description provided.

[silene]

None of the remaining failures are due to the added packages, so this is ready.

[haochenx]

@silene it's probably a good idea to up stream #23843

[silene]

The archive already contains a fully functional configure script; there is no reason to force the users to install autoconf to produce a new one.

[haochenx]

Then I guess maybe the dependency to conf-autoconf should be removed?

[silene]

No, it is needed for people who are directly building from the git repository, since we are not versioning the automatically generated configure script, hence why the dependency on conf-autoconf is only there when Opam runs in dev mode.

[haochenx]

Ah. So does it mean patches like #23843 is no longer needed for this new version?

[silene]

It was not needed either for the older releases, since the configure script has always been shipped in the archive. But for some unknown reason, in some very rare cases, the configure script was considered to be out-of-date, which caused Make to try to regenerate it from scratch. (A filesystem issue?) So, yes, depending on conf-autoconf does work around the issue, but it is way overkill. It is much cheaper to just ensure that configure is up-to-date by touching it just after extraction.

[haochenx]

Okay I see. So if my understanding is correct these are not necessary: I will dismiss the suggestions.

[haochenx]

All looks good to me. CI failures are unrelated as mentioned by the author

[mseri]

Thanks for the explanation. Should I backport this to the old why3 versions and reinstated conf-autoconf as a dev-only dependency?

[silene]

I do not think it is worth the effort, now that Why3 1.7 is available.