bug#36237: Support (rx (and (regexp EXPR) (regexp-quote EXPR)))

["Mattias Engdegård" <mattiase@acm.org>]

16 juni 2019 kl. 14.25 skrev Noam Postavsky <npostavs@gmail.com>:

>>> The plan was to replace rx with ry entirely when complete.
> 
> How far away is this?  Would it make sense to delay this bug until "ry"
> comes in?

The function-complete ry has been put at https://gitlab.com/mattiase/ry for the time being. It should now be entirely compatible, including support for `rx-constituents'. Your proposed `literal' was also added, which was instructive; I needed to know how it would fit in.

Bootstrapping Emacs with it (ry renamed to rx, of course) works, including python-mode which uses `rx-constituents' for extension. It also has fewer of the known bugs of the current rx, and a much more extensive test suite.

There are two things that perhaps need to be resolved before replacing the current rx:

1. Is there externally developed elisp code that makes use of internal rx functions and variables, perhaps for extension purposes, and do we need to worry about breaking it? The code is old enough not to delineate internal and public symbols clearly (no double-dash), which means that people may have gone and used all sort of things.

2. What would a good extension mechanism look like, and should one be put into place right away, so that we can point to a decent replacement for the internal toys that users relied upon?

Noam, unless the consensus is that ry is unequivocally as good or better than rx, you could just as well apply your patch (suitably fixed up). Even if later replaced, there is nothing fundamentally wrong with the design; let's not hold it hostage.