GNU bug report logs - #39452
[PATCH] vc-git-state fails for filenames with wildcards

Previous Next

Full log

View this message in rfc822 format

From: Eli Zaretskii <eliz <at> gnu.org>
To: Dmitry Gutov <dgutov <at> yandex.ru>
Cc: npostavs <at> gmail.com, 39452 <at> debbugs.gnu.org, Wolfgang.Scherer <at> gmx.de
Subject: bug#39452: [PATCH] vc-git-state fails for filenames with wildcards
Date: Fri, 14 Feb 2020 17:45:56 +0200

> Cc: npostavs <at> gmail.com, 39452 <at> debbugs.gnu.org, Wolfgang.Scherer <at> gmx.de
> From: Dmitry Gutov <dgutov <at> yandex.ru>
> Date: Fri, 14 Feb 2020 16:40:04 +0200
> 
> On 14.02.2020 16:14, Eli Zaretskii wrote:
> > Any command that prompts for a file name, I guess.  vc-delete-file and
> > vc-rename-file come to mind.
> 
> In both of these commands entering a non-trivial pathspec is both 
> undocumented and hard to do: look at the interactive form, it calls 
> read-file-name with MUSTMATCH t. In other words, it doesn't let you 
> input interactively anything that's not an existing file name.
> 
> > But my comment was more general: we don't plan on not supporting Git
> > specs in file names, do we?
> 
> I don't see how we'd keep supporting them in these two particular 
> commands without keeping bugs similar to this one unfixed (e.g. 'M-x 
> vc-delete-file test[56].xx' where test[56].xx is an existing filename).
> 
> AFAICS, they're working purely by accident. That's not to say we can't 
> introduce new versions of these commands that would accept pathspecs (or 
> do it with C-u, etc).

I won't argue.  I just wanted to point out that using Git signatures
internally might get in the way, whereas environment variables and
command-line switches are free from that disadvantage.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.