Re: [elpa] externals/standard-themes 0604883ecc: Expand 'deftheme' with metadata

[Philip Kaludercic]

Protesilaos Stavrou <info@protesilaos.com> writes:

>> From: Philip Kaludercic <philipk@posteo.net>
>> Date: Fri, 25 Aug 2023 07:43:55 +0000
>>
>> ELPA Syncer <elpasync@gnu.org> writes:
>>
>>> branch: externals/standard-themes
>>> commit 0604883ecc89f37f2d8fcb33ec3c8f7f2b8bbe2e
>>> Author: Protesilaos Stavrou <info@protesilaos.com>
>>> Commit: Protesilaos Stavrou <info@protesilaos.com>
>>>
>>>     Expand 'deftheme' with metadata
>>>     
>>>     This is to support new features in Emacs where themes can specify
>>>     the set they belong to, as well as whether they are light or dark.
>>>     The built-in command is 'theme-choose-variant'.
>>>     
>>>     This is in response to Emacs bug#65468:
>>>     <https://debbugs.gnu.org/cgi/bugreport.cgi?bug=65468>.  Thanks to
>>>     Mauro Aranda for bringing this matter to my attention.
>
>> [... 20 lines elided]
>>
>> Please note that this will cause an issue for anyone using the package
>> before Emacs 29, because deftheme only accepts 1-2 arguments before
>> da2e6da72296.
>
> Thank you Philip!  In that the case, I will have to revert the commit
> and only apply it to the modus-themes on the emacs.git trunk.
>
> That granted, I am evaluating the following with Emacs 28 and it does
> not throw an error (see attached screenshot as well):
>
>     (deftheme modus-operandi
>       "Elegant, highly legible theme with a white background.
>     Conforms with the highest legibility standard for color contrast
>     between background and foreground in any given piece of text,
>     which corresponds to a minimum contrast in relative luminance of
>     7:1 (WCAG AAA standard)."
>       :background-mode 'light
>       :kind 'color-scheme
>       :family 'modus)

My bad, the old definition of `deftheme' actually had a &rest _ignored
at the end of the argument list, but that was removed in a4a35305 (which
according to git tag --contains was after Emacs 28).  From what I
recall, in bug#57639 the issue was related to the scraping of autoload
cookies, and that before that report, the scraper would have copied the
entire definition, while the new definition just copies the properties.