IRC channel logs

<dthompson>nij-: the point was that guile could provide the platform on which emacs could run emacs lisp code, which would have given emacs better VM and compiler infrastructure, threading support (which emacs has now but didn't then), and would just generally allow emacs development to be more focused on editor features than language ones.

<nij->emacs has its own VM

<nij->the byte code machine

<nij->Does guile emacs reimplement a similar VM in itself?

<Arsen>well it uses guile

<nij->Or would guile be that VM in the project?

<nij->The semantics of Emacs' VM is fixed.

<nij->For elisp codes to work properly we need that old VM.

<Arsen>not really, it's not particularly hard to swap out that implementation detail

<nij->So I'd guess even if someone rewrites it in another nice lang (like guile), it still has to abide to the old VM's semantics.

<johnjaye>nij-: the former. guile is a big vm

<nij->Yes, johnjaye, I expect that.

<nij->Ok, then we need to keep that old VM's semantics.

<johnjaye>well then it's simple, in principle. you implement your language but on someone else's VM.

<johnjaye> turing tarpit.

<nij->What would be any benefit to have that old semantics replemented in guile?

<johnjaye>because you have the ability of the guile vm. like threading.

<johnjaye>it's probably like clojure on the java vm?

<Arsen>the emacs VM bytecode is an unstable implementation detail, so replacing it with the guile one isn't that difficult

<Arsen>as for language semantics, that doesn't depend on the actual VM implementation

<nij->i see, johnjaye , future development will be easier

<dthompson>at the time the project was active, the elisp implementation on the guile vm was more-or-less bug-for-bug compatible.

<johnjaye>in theory you could also write emacs extensions in scheme

<dthompson>there were performance issues. it was not a production-ready implementation. but it worked.

<dthompson>the emacs developers weren't really into it, though, so that was that.

<Arsen>heh, unfortunate

<dthompson>the technical problems could have been overcome if there was enthusiasm and buy-in. alas...

<johnjaye>sounds like the finger-pointing set in quickly after it faded out

<dthompson>from what I remember, it seems like the emacs perspective was that guile was being forced on them or something. I didn't really feel that was the case. I think guile people were certainly excited about the idea.

<dthompson>but whatever the specific reasons, the buy-in wasn't there. it's a bummer but it is what it is.

IRC channel logs

2023-05-01.log