Re: [PATCH] add compiled regexp primitive lisp object

[Danny McClanahan <dmcc2@hypnicjerk.ai>]

Thanks so much for this thorough and thoughtful response! I am compiling a 20-minute recorded talk for the upcoming EmacsConf regarding these discussions and investigations. I am also now waiting for a response from graduate school, during which I hope to then have more time to pin down several of the threads raised by this discussion.

This has all been immensely helpful and I really appreciate how responsive emacs-devel has been in elaborating much of the packed-together assumptions I was making.

On Tuesday, August 13th, 2024 at 21:32, Stefan Monnier <monnier@iro.umontreal.ca> wrote:

> 
> 
> > Ignoring the issue that regexps may be syntax-table dependent and other
> 
> > complications (issues which also would need to be addressed in the
> > proposed approach), what would be the advantage of exposing compiled
> > regexp objects versus having an infinite regexp cache?
> 
> 
> I can think of the following:
> 
> - Not having to pay as much attention to the engineering of the cache
> (it's currently small because the lookup could become costly, with
> the current implementation).
> 
> - We could offer to spend a fair bit more time optimizing the compiled
> code (e.g. removing backtracking).
> 
> But there's always the question of whether it's worth the complexity.
> IOW, what are the concrete use cases where it makes
> a measurable difference.
> 
> 
> Stefan