12:30 <neiluj> hi! is there a way for menhir to cache the parser it produces given a grammar?

12:51 <contificate> Does using dynamic_include in dune undermine incremental builds in any way?

12:52 <contificate> I've went to some effort to split-up an auto-generated module into many modules that are then include'd into the original module. Yet, when I do a small change, the hashes for almost all files remain the same after being generated again, but dune's tracing seems to show they're being rebuilt anyway

13:04 <discocaml> <otini_> well, the automaton produced by menhir is either a big table in memory or directly a set of recursive functions, so I’m not sure how more cached it could be

13:09 <contificate> What does menhir have to do with anything?

13:10 <discocaml> <darrenldl> (otini was replying to neiluj where they were asking about menhir stuff)

13:11 <contificate> Oh, sorry - I don't have history on IRC. I saw "cached" and assumed it was a response to me

13:12 <neiluj> thx otini! so one could store the automaton in a file and load it

13:12 <neiluj> neat!

13:14 <contificate> Doesn't Menhir execute pretty fast? Clever construction algorithms that are an extension of the classic LALR(1) algorithm - of which I think IELR is effectively that and state splitting - are extremely efficient.

13:16 <contificate> Unless you are talking about using Menhir as a library and wish to serialise what it outputs and then interpret it with the support library

15:13 <discocaml> <otini_> >thx otini! so one could store the automaton in a file and load it

15:15 <discocaml> <otini_> neiluj: well that's how menhir is usually used, it takes a .mly file and produces a .ml one which implements the parser (if you used Dune, that happens behind the scenes)

15:15 <discocaml> <otini_> contificate: TIL that dynamic_include exists

16:28 <discocaml> <contificate> `dynamic_include` is so neat