bug#62694: 30.0.50; eglot-tests fails with recent pylsp

[João Távora]

Basil Contovounesios <contovob@tcd.ie> writes:

> The CI probably installs anew each time it runs, whereas I was hinting
> (sorry for not making it clearer) at the case where a local user updates
> pip packages over time.  

Well, I'm a "local user" updating pip packages overtime, and used that
command and it worked (i used to to rev up pylsp to 1.7.2).  Anyway, we
were talking EMBA anyway, which is a CI system installing it anew each time.

>> Who knows? Did you try it?
>
> I only tried the aforementioned patch at the time.
>
> I was just clarifying for posterity that the issue may well have lain
> with a component of pylsp rather than the version of the server itself,
> which is what this discussion focused on.

Sure, could be.  Any number of things could have been it.  I've just
forcibly uninstalled autopep8 here and I just got one failure, not all
that Michael reported.  The one thing that could have solved it was
never even tried.

>> If it works i would have been happy to accept it Noone proposed
>> it. Instead, some truly vague and complex vapourware things were
>> proposed.
>
> That earlier proposals were not complete, optimal, or up to yours or
> anyone's standards does not warrant degrading them, IMO.  

- vague: there are no details, just the idea of "checking if the server
  is up to the job".
  
- complex: whatever its elusive nature is, probably more complex than
  (skip-unless (executable-find "pylsp"))

- vapourware: 0 code

So no degradation or even standard-checking going on here.  Primarily
because there was absolutely nothing brought forth to degrade or to
check standards on.

Also FTR, I wasn't the one who suggested I was doing other people's
homework.

> Like you, we're all here just trying to improve Emacs in whatever way
> we can.

Sure.  I believe thanked Michael multiple already for taking an interest
in Eglot's tests.  If I didn't, I do so again.  And I thank you too, of
course.  But it's certainly not "whatever way we can" if you obstinately
refuse to even try the one-liner suggestion of your interlocutor that
would probably fix the EMBA issue.

The academically postulated, never-demonstrated, "larger" issue of Emacs
developers running Eglot tests and facing non-determinism w.r.t to
external programs (why not w.r.t. cosmically induced bit-flipping, for
that matter?) probably cannot ever be fixed in this universe.

João