Re: support for git commit --amend/--signoff

[Stefan Monnier <monnier@iro.umontreal.ca>]

>> That doesn't bother me too much.  I wish we could move further away from
>> C-x v v since I find it doesn't really work for modern VCSes.
> Not sure why you think so, but please don't try to take it away, it
> has been in use for so long, and it's still very useful.

I guess "C-x v v" in itself is not a bad key and it does something
useful, so in this sense I don't intend to get rid of it, but its
binding to vc-next-action is not very useful, since in modern VCSes, the
only thing it can ever do (or close enough) is "commit", so the name
is misleading.

> Also --amend it's still a property of commit, it's not a different
> command, why make it a different command in emacs?

Yes, that's the main argument in favor of keeping it within the
"vc-commit" command (currently misnamed vc-next-action).

>>> and given the reluctance of some other VCs to change history, it does
>>> not seem that it will be generalized.
>> DaRCS already supports a more general form of Git's amend.  Bazaar is
> No idea, we don't support DaRCS at all...

We do want to support DaRCS, tho.

>>> More, it does not solve the problem of --signoff (which might be
>>> adopted by other VCs).
>> What's the problem with sign-off?  It seems this one fits perfectly well
>> inside a "Signed-Off-By:" header.
> Signed-Off-By: as an empty header does not work.  --signoff does not
> have any arguments.

Oh, that's why you tend to treat it like --amend.  I see it's a problem,
indeed, but I don't think that storing it into a buffer-local variable
is a good answer.  I'd rather have a "Sign-Off: yes" (you can still
have a vc-git-log-edit-toggle-signoff command that adds/removes such
a header).

I was thinking of (re)using the Signed-Off-By thingy used by Linux kernel
developers, but I'm not exactly sure of how it's intended to be used, so
I'm not completely clear if and how it could be cleanly integrated with
VC's need to support --signoff.

>> Why add a var "log-edit-get-extra-flags-function" that's never used by
>> log-edit?
> It's used by vc-checkin, the main user of log-edit and by the modes
> derived from log-edit.  Do you have a better proposal?

Yes: use a name that starts with "vc-" since it's a variable added for
VC, and used exclusively by VC.

>> Also as a user I'd *really* like to have a clear visual feedback about
>> the fact that I'm amending something, which is another reason why having
>> it inside the headers is an attractive direction.
> For that in my local tree I have a minor mode instead of
> vc-git-log-edit-toggle-amend, it shows "amend" in the modeline.

Why not show it inside the buffer (e.g. in the header ;-) instead?


        Stefan