GNU bug report logs - #70036
30.0.50; Move file-truename to the C level

Previous Next

Full log

Message #230 received at 70036 <at> debbugs.gnu.org (full text, mbox):

From: João Távora <joaotavora <at> gmail.com>
To: Theodor Thornhill <theo <at> thornhill.no>
Cc: Eli Zaretskii <eliz <at> gnu.org>, 70036 <at> debbugs.gnu.org,
 felician.nemeth <at> gmail.com
Subject: Re: bug#70036 a fix that
Date: Fri, 19 Apr 2024 12:40:06 +0100

On Fri, Apr 19, 2024 at 12:32 PM Theodor Thornhill <theo <at> thornhill.no> wrote:

> I wonder - why can't we use just buffer-file-truename?
>
> In the below function, wouldn't buffer-file-truename let us avoid the
> file-truename in eglot-path-to-uri?

Never used, but try it out.  It seems it's already a built-in cache
of whatever the buffer's truename is (and the docstring says
it does call the slow `file-truename`).  But like in my patch, it's
a one-time thing.  So yeah, if it's available in old Emacsen as well,
 it could replace `eglot--cached-tdi`, but then the search in the
publishDiagnostics
handler would have to be adjusted accordingly to call uri-to-path again.

Try it out, may be cleaner indeed.

Beware though, Stefan has fixed the tests which is very good, but there
still isn't a test for the symlink scenario, so you'd have to test it
manually (or
even better, craft that test yourself, with clangd).

João

Previous Next

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