Re: [PATCH] contrib: remote-helpers: add move warnings (v2.0)

Felipe Contreras <felipe.contreras@gmail.com> writes:

You seem not to understand at all what a regression is.

My understanding is that versions of remote-hg shipped with all versions of Git did not work with Hg 3.0, so not working with Hg 3.0 is a regression in v2.0 at all. A recent report was about Hg 3.0 not working with 1.9.3, but I think you earlier said all versions of Git does not work with Hg 3.0, and I can believe it. That is hardly a regression.

You could argue that Hg has a new regression to its external users of its API when it went to 3.0. We actually had a similar breakage in 1.5.4, where it was reported late in the cycle after -rc0 [*1*] that cgit that linked with our internal API libgit.a was broken by a change on our side, which resulted in us fixing the breakage (even though technically you may be able to say that it was cgit's fault to link with libgit.a in the first place) with 18125644 (Move sha1_file_to_archive into libgit, 2008-01-14) very late in the cycle. Calling that a regression in cgit would have been insane, even if we did not patch our side up to accomodate it.

Stop this idiocy.

[References]

*1* http://thread.gmane.org/gmane.comp.version-control.git/70117/focus=71064