Re: In support of Jonas Bernoulli's Magit

[Eli Zaretskii]

> From: Marcin Borkowski <address@hidden>
> Date: Sun, 09 Jul 2017 11:25:28 +0200
> Cc: address@hidden, Richard Stallman <address@hidden>,
>       Stefan Monnier <address@hidden>
> 
> The main annoyance is that if I edit _more than one file_, and press C-x
> v v, instead of committing my all changes (or at least asking me about
> it), it only commits the changes in the currently edited file.  Several
> times I made an incomplete commit because of that.  I tried to find an
> option to override this, but to no avail.

You can do what you want with VC by first marking the files you want
to commit in vc-dir buffer.

> I guess that Emacs' VC with distributed VCSs is fundamentally broken,
> since it was really designed with RCS and its likes in mind

You seem to talk only about "C-x v v".  That command doesn't assume
RCS-type VCS, it assumes that there's some more-or-less constant
sequence of actions in your workflow, and attempts to support that
sequence by doing "the next thing".  I think this concept can be
naturally generalized to modern VCSes, except that the number of
different workflows is larger, definitely more than one.  At the time,
I proposed to move in that direction, but people disagreed, so nothing
happened.  I still think it's a good idea, so maybe someone will want
to work on it.

But VC includes more than "C-x v v" alone.  There are commands like
"C-x v D", "C-x v ~" (just recently mentioned as very important and
useful -- with Git, no less!), "C-x v g" with its sub-commands,
"C-x v l", "C-x v L", "C-x v I"/"C-x v O", "V-x v u", and others.  I
see no reasons to ignore these useful commands when talking about VC.

> Sorry if that sounds harsh, but I consider VC's utility (with DVCSs) to
> be zero _at best_.

I think this is at least exaggerated.  I certainly don't see how this
could follow from what VC provides.