Re: [NonGNU ELPA] new package: eglot-inactive-regions

[Philip Kaludercic <philipk@posteo.net>]

Filippo Argiolas <filippo.argiolas@gmail.com> writes:

> Philip Kaludercic <philipk@posteo.net> writes:
>
>> Filippo Argiolas <filippo.argiolas@gmail.com> writes:
>>
>>> Philip Kaludercic <philipk@posteo.net> writes:
>>>
>>>> Filippo Argiolas <filippo.argiolas@gmail.com> writes:
>>>>
>>>>> Philip Kaludercic <philipk@posteo.net> writes:
>>>>>
>>>>>> Filippo Argiolas <filippo.argiolas@gmail.com> writes:
>>>>>>
>>>>>>> Filippo Argiolas <filippo.argiolas@gmail.com> writes:
>>>>>>>
>>>>>>>> Philip Kaludercic <philipk@posteo.net> writes:
>>>>>>>>
>>>>>>>>> I would try something of the form like
>>>>>>>>>
>>>>>>>>>   (if (fboundp 'new-function)
>>>>>>>>>       (new-function ...)
>>>>>>>>>     (old-function ...))
>>>>>>>>>
>>>>>>>>> If on the other hand there has already been a new release of Eglot with
>>>>>>>>> these commands, then just depend on that version and the issue would
>>>>>>>>> resolve itself.
>>>>>>>>>
>>>>>>>>
>>>>>>>> CC-ing João.
>>>>>>>>
>>>>>>>> I pushed the suggested change but I still get the compile warnings about
>>>>>>>> deprecated functions. I'd like to keep supporting emacs-29, should I
>>>>>>>> just disregard the warning?
>>>>>>>>
>>>>>>>> Sorry for the ignorance, how would the dependency on a specific eglot
>>>>>>>> version work now that it's in core?
>>>>>>>> Does it require the users to have additional repos enabled?
>>>>>>>>
>>>>>>>> Sounds strange to me that we have macros to deprecate functions but no
>>>>>>>> mechanism to automatically switch to the new one if it's just a
>>>>>>>> rename. Am I missing something obvious?
>>>>>>>>
>>>>>>>> Thanks!
>>>>>>>>
>>>>>>>> Filippo
>>>>>>>
>>>>>>> How about something like:
>>>>>>>
>>>>>>>   ;; fallback to deprecated eglot functions
>>>>>>>   (when (version< emacs-version "30")
>>>>>>>     (defalias 'eglot-uri-to-path 'eglot--uri-to-path)
>>>>>>>     (defalias 'eglot-range-region 'eglot--range-region))
>>>>>>>
>>>>>>> or even something similar with a check on eglot version?
>>>>>>
>>>>>> That can be dangerous if other packages do fboundp checks and infer too
>>>>>> much from that.  I would try to see if adding `declare-function's could
>>>>>> help suppress the warnings?
>>>>>
>>>>> Nope, the only solution working so far is to call the deprecated
>>>>> functions with `with-no-warnings'.
>>>>>
>>>>> Guess I'll go with this. I'd be fine with the warnings too if we didn't
>>>>> have the annoying habit of scaring the end users with them :-)
>>>>
>>>> I don't have your source code available right now, my last suggestion
>>>> would be to use `with-suppressed-warnings' where  you can suppress
>>>> specific obsoletion warnings, but that requires Emacs 27.1 or newer.
>>>>
>>>
>>> Thanks this seems to do the trick!
>>
>> OK great, I'll add the package to NonGNU ELPA.  Can you just add a
>> .elpaignore file that lists the "screenshot" directory to be excluded
>> from the tarball?
>>
>> PS. Using the command "oxipng -o max -Z" I could compress one of the
>> images by 49%.  You might also consider creating a SVG screenshot.
>
> Great, thanks!  Added screenshots to ignore, let me know if I also have
> to bump the release number.

That will be necessary for the change to take effect.

> One of these days I wanted to update screenshots as they look ugly on
> lower res displays, will consider switching to SVG ones.

In case you or anyone else reading along is not familiar with the
function, you can generate a SVG screenshot by evaluating:

  (write-region (x-export-frames nil 'svg) nil "screenshot.svg")

Note that this requires Emacs to have been built using Cairo.