bug#68028: 29.1; (elisp) `pcase Macro': move `rx' before SEQPAT description

[Mattias Engdegård]

27 dec. 2023 kl. 22.47 skrev Stefan Kangas <stefankangas@gmail.com>:

>> Please consider moving the description of `rx' forms from after the
>> description of SEQPAT forms (`and' and `or').  The scope of the latter
>> description is less clear if `rx' follows the descriptions of `and' and
>> `or' (with no particular separation).  IOW, separate the SEQPAT
>> description from the other pattern descriptions better.
> 
> Mattias, since you added that part, do you have any thoughts?

The location of the `rx` pattern description is fine; it's the SEQPAT paragraph 
that causes trouble as it splits the table into two parts.

I suggest we remove that paragraph because it doesn't actually say anything 
useful at that point; the user can learn how `or` and `and` patterns work 
without knowing that they are SEQPATs. Actually, I'd hoist those two patterns 
and reorder them to keep patterns in a rough order of usage: something like

  pred, guard, or, and, let, app, rx

makes more sense -- `pred` and `guard` clearly belong together, `app` is less 
common than `let`.

We could explain SEQPAT in a short sentence somewhere else, either near the 
beginning of the node or further down in the 'Caveats' subsection. A simple 
(and probably flawed) patch attached.

Stefan M probably has a better sense of how to improve the text. (By the way, 
the caveat section says that sequencing patterns use `eq` for comparison. Don't 
they use `eql`?)

seqpat.diff
Description: Binary data