bug#43117: [PATCH] Add .git-blame-ignore-revs file

[Eli Zaretskii]

> From: Stefan Kangas <stefan@marxist.se>
> Date: Mon, 19 Oct 2020 16:06:45 +0000
> Cc: leungbk@mailfence.com, larsi@gnus.org, rpluim@gmail.com, 
>       43117@debbugs.gnu.org, dgutov@yandex.ru
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > Why would someone want to skip commits in "git blame"?  We need to
> > have agreed-upon criteria or policy for that, otherwise each one of
> > use will step on the others' toes.  "git blame" is an important
> > forensics command, so hiding commits from it might surprise someone,
> > if it's unexpected.
> 
> For example, if a commit fixed a typo in a doc string, I'd much rather
> see the commit that added that text than the one that fixed the typo.

But that's your personal preference, isn't it?  Do we know for sure no
one will ever want to see changes that fixed typos?  For example, what
about some "typo" that is controversial (like the "parseable" thing we
discussed lately), and we want to know who and why "fixed" it?

In short, the more I think about this feature, the less it makes sense
to me to use it in our project.  It's probably fine for a
single-developer projects, where the preferences are never in
conflict.  But ours is a very different project.

> >> Wouldn't having non-existent commit ids in there mostly be an aesthetic
> >> problem?
> >
> > I don't know.  Will it?  I have no experience with this feature.
> 
> AFAIU, git will just skip over non-existent commits in this case.

It requires a setting in .git/config, doesn't it?  Would older
versions of Git warn about the setting they don't know about?  Or do
we expect users to turn this on in their ~/.gitconfig instead?