Restore emacs-29.1 eglot behaviour on master

Restore emacs-29.1 eglot behaviour on master

[Pedro A. Aranda]

Hi

I've been having massive problems with eglot on master. I understand 
evolution, being on master, etc. but this is taking too much time now 
<sigh> and some LSP implementers understand but others don't even react...

I'm mainly working with Python and LaTeX and this is my eglot config:

   (setq-default eglot-workspace-configuration
                 '(:pylsp
                   (:configurationSources ["pycodestyle"]
                    :plugins (:autopep8 (:enabled t)
                              :jedi_completion (:include_params t
                                                :eager t
                                                :fuzzy :json-false)
                              :flake8      (:enabled :json-false)
                              :pylint      (:enabled :json-false)
                              :pycodestyle (:enabled :json-false)
                              :pyflakes    (:enabled t)
                              :mccabe      (:enabled t ;; boolean: true 
(default) or false
                                            :threshold 45)))
                   :texlab
                   (:completion (:matcher "prefix"))))

What I have learnt:

On emacs-29, the section relating to the LSP that is used when on a 
specific buffer (and only that) is sent tp the LSP. And everything woks 
as expected

On emacs-30. the full configuration is sent. That confuses the LSP 
servers and they start working with the default configuration, which is 
sometimes of little help.

Could the original behaviour be (re)implemented in master, please

Thanks a ton in advance,

/Pedro A. Aranda

Re: Restore emacs-29.1 eglot behaviour on master

[Dmitry Gutov]

On 18/02/2024 08:36, Pedro A. Aranda wrote:
> On emacs-29, the section relating to the LSP that is used when on a 
> specific buffer (and only that) is sent tp the LSP. And everything woks 
> as expected
> 
> On emacs-30. the full configuration is sent. That confuses the LSP 
> servers and they start working with the default configuration, which is 
> sometimes of little help.

This seems like a good material for a bug report.

Re: Restore emacs-29.1 eglot behaviour on master

[Eli Zaretskii]

> Date: Sun, 18 Feb 2024 07:36:46 +0100
> From: "Pedro A. Aranda" <paaguti@gmail.com>
> 
> Hi
> 
> I've been having massive problems with eglot on master. I understand 
> evolution, being on master, etc. but this is taking too much time now 
> <sigh> and some LSP implementers understand but others don't even react...
> 
> I'm mainly working with Python and LaTeX and this is my eglot config:
> 
>    (setq-default eglot-workspace-configuration
>                  '(:pylsp
>                    (:configurationSources ["pycodestyle"]
>                     :plugins (:autopep8 (:enabled t)
>                               :jedi_completion (:include_params t
>                                                 :eager t
>                                                 :fuzzy :json-false)
>                               :flake8      (:enabled :json-false)
>                               :pylint      (:enabled :json-false)
>                               :pycodestyle (:enabled :json-false)
>                               :pyflakes    (:enabled t)
>                               :mccabe      (:enabled t ;; boolean: true 
> (default) or false
>                                             :threshold 45)))
>                    :texlab
>                    (:completion (:matcher "prefix"))))
> 
> What I have learnt:
> 
> On emacs-29, the section relating to the LSP that is used when on a 
> specific buffer (and only that) is sent tp the LSP. And everything woks 
> as expected
> 
> On emacs-30. the full configuration is sent. That confuses the LSP 
> servers and they start working with the default configuration, which is 
> sometimes of little help.
> 
> Could the original behaviour be (re)implemented in master, please
> 
> Thanks a ton in advance,

João, any comments or suggestions?

Thanks.

Re: Restore emacs-29.1 eglot behaviour on master

[João Távora]

On Thu, Feb 22, 2024 at 1:39 PM Eli Zaretskii <eliz@gnu.org> wrote:

> João, any comments or suggestions?

Not many.

1.Pedro seems to be using per-project configuration as user
configuration. It's best to use other methods of user configuration.
See manual.
2. exactly when or how this "sending" happens is missing. There is no
recipe, and there should probably be one.  Maybe in a bug report
following the instructions in the manual?

I'm unsubscribed from this list and away from Eglot/Emacs development
atm so I can't be of more help, sorry.

João