João Távora <joaotavora@gmail.com> writes:

> Stefan Monnier <monnier@iro.umontreal.ca> writes:

> Does that "some" include _this_ problem?  Let's take this one for
> example.
>
>    make -C test filenotify-tests
>    SELECTOR=file-notify-test04-autorevert

Sorry, here I meant to type 'file-notify-test04-autorevert-remote'

> We know this was introduced by the recent changes
> 54ef338ba3670415cf47fabc33a92d4904707c7e.

FWIW, I've run the following experiment:

1. Reverted 54ef338ba3670415cf47fabc33a92d4904707c7e

2. Applied a simpler, more conservative, patch, discussed with Michael
   before.  The patch keeps the protocol for tramp-a-p-o untouched.  It
   simply adds a "while" to the 'a-p-o' call issued for "related"
   processes.

3. Ran the filenotify test suite.  All but one tests pass.
   'file-notify-test04-autorevert-remote' now passes.  The failure I
   get, 'file-notify-test06-dir-validity-remote', is _also_ in Emacs 29

4. Removed the Tramp-specific Eglot workaround in
   lisp/progmodes/eglot.el which disabled Controlmaster.

5. Ran the Eglot-over-Tramp experiment with the JDTLS LSP server
   multiple times with 0% failures.

6. The two Tramp-related failures in the eglot-tests.el test suite are
   _also_ fixed.  Ran this multiple times, 0% failures.

It seems evident that my simpler patch puts us in a much better place
than we are now, so I propose we push them.  Patches are attached.

Of course, we can continue to explore other more elegant solutions.

João