GNU bug report logs - #76313
New function `replace-region`

Previous Next

Full log

View this message in rfc822 format

From: Tassilo Horn <tsdh <at> gnu.org>
To: Stefan Kangas <stefankangas <at> gmail.com>
Cc: Philipp Stephani <phst <at> google.com>, 76313 <at> debbugs.gnu.org, Eli Zaretskii <eliz <at> gnu.org>, Stefan Monnier <monnier <at> iro.umontreal.ca>
Subject: bug#76313: New function `replace-region`
Date: Thu, 20 Feb 2025 08:03:02 +0100

Stefan Kangas <stefankangas <at> gmail.com> writes:

>> The benefits of supporting the `replace-region-contents` API are much
>> smaller than the "sounded minor" annoyances we'd have to pay for the
>> rest of Emacs's future.
>
> AFAIU, the `replace-buffer-contents` API exists because:
>
>> There are many tools (e.g. auto-formatters) that take buffer
>> contents, reformat them, and write the reformatted output somewhere.
>> However, there is no good way how to apply the modified output to the
>> source buffer.  The naive way (erasing and re-inserting the buffer
>> contents) loses point and markers.  Therefore there should be a
>> function (e.g. `replace-buffer-contents') that calculates a minimal
>> diff between the old and the new contents and uses editing operations
>> (insert, delete, etc.) to apply the diff.  Ideally this would be done
>> without any external tools (e.g. 'diff').
>   https://debbugs.gnu.org/cgi/bugreport.cgi?bug=25355

That's true.  I use it frequently (through json-pretty-print-buffer) to
format minimized JSON replies from REST-APIs.  It would be a bummer if
point moved to the end of the buffer contents after formatting.  But it
can be unbearable slow if the region is very large and the replacement
text differs a lot from the current text.  That's why it has these
strange MAX-SECS and MAX-COSTS arguments.

> This sounds quite specialized to my mind, and the API is also quite
> complicated.  I think it's a fine addition for this use case, but I'd
> suggest renaming it to something that highlights its specialized
> nature.
>
> The `replace-region-contents' function builds on that, and similarly
> has a complicated calling convention.  I think it should similarly be
> renamed to highlight its specialized nature.

Fine with me but we have to consider that it's already used in the wild.
I remember submitting patches to at least restclient.el (which only uses
it through json-pretty-print-buffer so wouldn't be affected by renaming)
and hindent.el, see
https://github.com/mihaimaruseac/hindent/blob/master/elisp/hindent.el#L220.
Maybe there are other callers in third-party packages.

> Meanwhile, according to what we have learned in Bug#71370, many users
> are currently using this:
>
>     (setf (buffer-substring START END) "foo")

That's actually quite elegant (although it gives an obsolecence
warning).  It never occured to me that (buffer-substring START END) was
a generalized variable.

> I think this shows the need for a more low-level operation, and
> `replace-region` seems to fit this use case exactly.  So I would be in
> favor of installing it.

IMHO, it should at least reference replace-region-contents from its
docstring and explain the difference (and vice-versa).

Bye,
Tassilo

Previous Next

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