REPL, part 1: Added interpreter mode to compiler interface, interpreter parsing functionality by alexreg · Pull Request #64648 · rust-lang/rust (original) (raw)
I think that having an idea of the bigger target is exactly the kind of thing we should talk about, but unfortunately I don't really have the time to read a branch and try to "reverse engineer" what the larger design is from that. What I would like to see is probably a more elaborated version of the "summary" from this PR, that tries to show how the pieces will fit together in the end.
That's fair. And I didn't expect you to do as much. I just wrote this to say "hey, the branch is there in case you're curious". What I can do is write up an expanded summary doc for this PR and the upcoming ones, and the overall vision. On HackMD or something.
I'm not sure yet overall whether I want to land support for a REPL. I think a REPL would be a great thing. But we have a lot of stuff going on and a lot of priorities. So I guess it's partly a question of how "invasive" the support is, and what kind of ongoing support it seems likely to require. My impression from this PR is that it should be relatively minimal? This seems like the kind of information that we would be looking for in evaluating the design doc.
I can understand why, but I think if the intrusion into the rustc compiler is minimal (as I've planned it to be), then why not? I'm happy to take on "maintenance duties" as far as the few REPL-only bits of rustc are concerned. I'm convinced this PR is pretty unobtrusive, so I'm glad you agree there so far.
(I'd also like the design doc to be something that could move to rustc-guide to help people in understanding the changes that are being done. I personally find it really hard to review commits when I don't have a reasonable up-front idea of what the PR is trying to do.)
Yeah, I'll keep that in mind when writing it. If it's on HackMD, then others like you can add notes and adjust things here and there if you fancy... no obligation though.
@alexreg to be clear, the design meeting process is pretty lightweight! Just file an issue on the compiler-team repo. We can also schedule it at a time that's not the normal slot, if that slot doesn't work well for you.
Ah, didn't know, but sounds good. If we could somehow schedule a short one next week (giving me time to prepare this doc, and people to see it, but not so longer as to let this PR linger more or even fester), then perfect.