Re: Eglot tests fail on emba

Hi Michael,

I've had a look at the log file and I can't make any sense of it either.
Given you say tests pass on emacs-31 branch, here's what I can say:

* There haven't been any recent commits to Eglot.

* The differences between emacs and emacs-31 versions of
lisp/progmodes/eglot.el cannot be responsible for this.  All they do
is fix a docstring and change some versioning constants in defcustoms.
You can check yourself with git diff emacs-31..master lisp/progmodes/eglot.el

* There are 0 differences in test/lisp/progmodes/eglot-tests.el

* There are 0 differences for lisp/jsonrpc.el (a critical supporting library
for Eglot).

So that said, I'm inclined to the following conjectures:

* You were lucky/unlucky: the tests are instable in both versions and things
just happened to go sour on master, not on emacs-31

* Some different fixture setup (different rust-analyzer version) on master
and emacs-31 runs.

* Some underlying child process bookkeeping machinery has subtly changed
recently. Last change to src/process.c is from June though (and you authored it)
and looks innocent enough.

So, maybe someone can shed some light. I can't; not at the moment
at least.

João




On Sun, Aug 23, 2026 at 10:33 AM Michael Albinus <[email protected]> wrote:
>
> Hi,
>
> since one or two days, eglot tests fail on emba. There is no FAILED test
> being reported; the test simply stops with
>
> --8<---------------cut here---------------start------------->8---
> error in process sentinel: Process EGLOT (on-type-formatting-project/(rust-mode)) not running: finished
> --8<---------------cut here---------------end--------------->8---
>
> I have no idea which commit is related. Eglot tests in the emacs-31
> branch still work. Logfile appended.
>
> Best regards, Michael.
>


-- 
João Távora