Re: erc-ircfmt.el: Transient-based formatting for Erc (Emacs 29+)

[J.P.]

Alcor <alcor@tilde.club> writes:

>> Not sure if this is already supported by your patch, but it might be
>> nice to optionally restore control chars when copying and yanking an
>> already inserted message that contains `{fg,bg}:erc-color-faceN' faces.
>> Actually, long ago, ERC shipped with something attempting to provide
>> such functionality. But it was removed for reasons unknown (to me).
>
> Agreed. Adding C-w and M-w suffixes to the transient prefix is a good
> idea. Will try that out soonish, since I've found myself wanting
> something similar a few times.
>
> Tangentially related, but: In general, copying/yanking formatted text
> into the IRC prompt is sort of misleading (in a visual sense). Since
> copying/yanking a region also copies all text properties as well, it may
> *seem* (to a novice user) like the sent text would retain its
> formatting, even though this is not the case. I wonder if it would make
> more sense to strip any face properties on text inserted at the Erc prompt?

FWIW, when `fill-wrap' is active, it's supposed to remove select
properties via `erc-stamp--display-margin-mode':

  (add-function :filter-return (local 'filter-buffer-substring-function)
                 #'erc--remove-text-properties)

Perhaps we could promote that to baseline `erc-mode' copying behavior,
and then allow various modules to extend it somehow, e.g., by overriding
`erc--own-property-names' locally. For stripping face props when yanking
into the prompt area, maybe see `yank-transform-functions' (29+).

>>> If desirable, it might be possible to
>>> include this code (or something similar) in erc-goodies.el, so Erc can
>>> provide this functionality OOTB.
>>
>> If this is to find a home in the ERC tree, you'll need to open a bug
>> report for it and also ensure it doesn't error when compiling or loading
>> on Emacs 27. This last part wouldn't be necessary if we were to add
>> transient.el as a hard ELPA dependency, but I'm not sure we're ready to
>> make that decision anytime soon.
>
> It's possible to have it fallback onto a normal prefixed command tree if
> transient.el is not available. But I'm wondering what's the support
> matrix for Erc vis-a-vis Emacs releases. Does Erc target the last 3
> major releases? (i.e. 27.x, 28.x, 29.x), or is >=27 a fixed constraint?

Last three, AFAIK. But when it comes to 27 in particular there's a sunk
cost component at play in terms of attention paid, at least in 5.6. But
I don't think a fallback is really necessary for new features, in
general: users on older Emacsen should probably expect a somewhat
degraded experience.