GNU bug report logs - #73496
30.0.91; Eglot (and clangd) out of sync after switching git branches in magit

Previous Next

Full log

Message #17 received at 73496 <at> debbugs.gnu.org (full text, mbox):

From: Mikael Springer <mikael.springer <at> protonmail.com>
To: João Távora <joaotavora <at> gmail.com>,
 Eli Zaretskii <eliz <at> gnu.org>
Cc: 73496 <at> debbugs.gnu.org
Subject: Re: bug#73496: 30.0.91;
 Eglot (and clangd) out of sync after switching git branches in magit
Date: Thu, 26 Sep 2024 11:07:43 +0000

You are right, I have eglot-ensure enabled for c++ts-mode.

I'll try to create an emacs -Q recipe for this as soon as I can. I also 
agree and suspect that any outside file operation will work as a trigger.

/Micke.


On 2024-09-26 12:25, João Távora wrote:
> On Thu, Sep 26, 2024 at 11:08 AM Eli Zaretskii <eliz <at> gnu.org> wrote:
>>> Date: Thu, 26 Sep 2024 08:20:52 +0000
>>> From:  Mikael Springer via "Bug reports for GNU Emacs,
>>>   the Swiss army knife of text editors" <bug-gnu-emacs <at> gnu.org>
>>>
>>>
>>> 1. Visit buffer with C++ code -> Eglot starts.
>>> 2. Listing references for a variable in Eglot works.
>>> 3. Switch Git branch in magit.
>>> 4. Go back to C++ buffer. Listing references for a variable using Eglot
>>> does not work, error message in echo buffer; "jsonrpc-request:
>>> jsonrpc-error: "request id=44 failed:", (jsonrpc-error-code . -32602),
>>> (jsonrpc-error-message . "trying to get AST for non-added document"),
>>> (jsonrpc-error-data)"
>>> 5. Closing the buffer and opening it again fixes the problem, or doing
>>> eglot-reconnect also fixes the problem, listing references is working
>>> again.
>>> 6. This is true for all open C++ buffers handled by Eglot before the
>>> branch switch.
>> CC'ing João.
> It would be great if Mikael could make this into a Emacs -Q recipe
> to ease investigation and rule out interference from configuration etc.
>
> For example this isn't  true
>
>    Visit buffer with C++ code -> Eglot starts.
>
> unless you use eglot-ensure in your config.  And possibly Magit isn't
> needed to trigger this, a simple Git branch change on the side will do (or
> maybe it won't in which case Magit has to be investigated, too).
>
> A concrete example toy repo with, say, a single file would also
> be useful.
>
> João.

Previous Next

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