GNU bug report logs - #59693
29.0.50; treesitter in base buffer doesn't respond to modifications in indirect buffer correctly

Previous Next

Full log

Message #32 received at 59693 <at> debbugs.gnu.org (full text, mbox):

From: Stefan Monnier <monnier <at> iro.umontreal.ca>
To: Yuan Fu <casouri <at> gmail.com>
Cc: Eli Zaretskii <eliz <at> gnu.org>, 59693 <at> debbugs.gnu.org, miha <at> kamnitnik.top
Subject: Re: bug#59693: 29.0.50; treesitter in base buffer doesn't respond
 to modifications in indirect buffer correctly
Date: Sun, 04 Dec 2022 22:49:14 -0500

> Actually there’s a little bit of problem. When we edit the base buffer, we
> would want to update the parsers in all of its indirect buffers as well, and
> AFAICT there is no pointer from base buffer to the indirect buffer, only the
> other way around. 

Background note: I consider indirect buffers an attractive nuisance
and for this reason I think we should spend as little time as possible
improving them.  Instead we should encourage people to develop
alternative approaches.
I know we have a large number of bugs lurking in that area.

E.g.:

    Emacs -Q
    M-x clone-indirect-buffer RET
    M-x fundamental-mode RET

we see that `*scratch*` has lost its font-lock highlighting even though we
changed major mode in the "other" buffer.
Now go to `*scratch*` type some text: you see that font-lock properly
updates the new code's highlighting.
Then go to `*scratch*<2>` and type some more text: you should see that
font-lock does *not* properly update the new code's highlighting in the
base buffer.

AFAICT this is the exact same bug as what you're seeing, just with
font-lock instead of tree-sitter.  [ Of course, for font-lock it's worse
because font-lock uses text-properties (which are shared between the
base and the its indirect buffers) so it simply can't do its job
properly in indirect buffers.  ]

This bug has been with us since indirect buffers were invented, so
it has very low priority, AFAIC.

> Then base buffer can update all indirect buffers’ parsers, and indirect
> buffer can find its base buffer and update all the parsers, including the
> base buffer’s parsers and other indirect buffers’ parsers.

Don't bother, please.
Instead, I recommend you disallow the use of tree sitter in indirect
buffers.  And we should probably try and change `insdel.c` so it always
runs the `after/before-change-functions` in the base buffer (this should
fix the worst part of the problems).


        Stefan

Previous Next

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