GNU bug report logs - #54001
29.0.50; abbreviate-file-name has side-effects

Previous Next

Full log

View this message in rfc822 format

From: Jim Porter <jporterbugs <at> gmail.com>
To: Stefan Monnier <monnier <at> iro.umontreal.ca>, 54001 <at> debbugs.gnu.org
Cc: Daniel Mendler <mail <at> daniel-mendler.de>
Subject: bug#54001: 29.0.50; abbreviate-file-name has side-effects
Date: Mon, 14 Feb 2022 09:52:32 -0800

On 2/14/2022 9:19 AM, Stefan Monnier via Bug reports for GNU Emacs, the 
Swiss army knife of text editors wrote:
> Package: Emacs
> Version: 29.0.50
> 
> 
> Since:
> 
>      commit bf505a63f98ed61934a8fb81ec65c96859606b6e
>      Author: Jim Porter <jporterbugs <at> gmail.com>
>      Date:   Mon Nov 15 13:33:07 2021 +0100
> 
>      Support abbreviating home directory of Tramp filenames
> 
> `abbreviate-file-name` has significantly changed in its behavior:
> - it's slower (because it goes through file-name-handlers)
> - it can have very visible side effects like prompting the user for a password.
> 
> I haven't measured the slowdown, so I'll assume it's acceptable, but
> asking for a password (or contacting a remote host) is not.

Sorry about that. I did what I could to minimize the slowdown (including 
some more general optimizations to make Tramp faster). There are some 
benchmarks in the original bug here (these are with 1000 iterations; 
you'll want to compare the first section with the last): 
<https://lists.gnu.org/archive/html/bug-gnu-emacs/2021-11/msg01293.html>.

> I suggest we take a step back and think of how to get that feature
> without having to contact any remote host during `abbreviate-file-name`.
> Maybe we can do that by making Tramp opportunistically add entries to
> `directory-abbrev-alist` when it performs expansion?

I think Michael Albinus suggested doing that in the original bug, 
although I was concerned about modifying defcustoms invisibly like that. 
Is that ok to do? Another option might be to store the abbreviations for 
a given file-name-handler somewhere internally and consult that when 
calling that file-name-handler's implementation of `abbreviate-file-name'.

Maybe this patch should be backed out for now; it shouldn't be 
interrupting the user. (I thought I'd tested that, but maybe it was on 
an earlier revision of the patch.) I'll probably have time to look into 
a new solution in a few weeks, but anyone else who's interested should 
feel free to fix it in the meantime.

Previous Next

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