Closed nikomatsakis closed 2 years ago
as one who is less familiar with the lang-agenda goings-on, I’m not 100% certain what you mean by registered comments.
would you be willing to clarify?
Given the following description, assuming it’s accurate, would I be correct to assume you mean the “preceding summary comments” that come before the FCP (described in the second sub-bullet)?
Ah, I somewhat mistyped. I meant registered concerns. But really I think what I would want to do is to reproduce (a) the summary comment and (b) the content of the FCP comment (which includes a set of checkboxes along with a set of registered concerns).
So as an example, reproducing the content of this comment would be a start; ideally we'd include this comment too to give more context.
FWIW, having looked a little into this, it will likely require modifications to the rfcbot API -- it currently only exposes the Markdown of the FCP comment. Maybe it's enough to inline those, but IIRC one of the additional problems is that there's no way to ask the rfcbot API for issues which are under FCP for a particular team (though that information is presumably tracked somehow -- it's displayed on rfcbot.rs). I'm capable of reviewing and deploying changes there, though; see https://github.com/rust-lang/rfcbot-rs/ for the source code.
I'm significantly less familiar with rfcbot than triagebot, so am not really able to point in the right direction though.
The API is https://rfcbot.rs/api/all, by the way -- it technically has the barebones of the right information (e.g., you can pull the comments Niko is looking for from it for sure), but I'm not sure if it's easy to pull everything from it. Maybe starting with what it has is a good first step though -- it's probably easier than doing upstream modifications first :)
@Mark-Simulacrum we know the set of issues already based on github labels, can we ask rfcbot for the markdown test for a particular issue? that'd be a good start.
Not directly, but I think the rfcbot API does contain enough to find it from the full set -- probably a good approach.
Thanks for the clarifications, y'all. Familiarizing myself with this now.
Here's my current understanding.
fcp.fk_bot_tracking_comment
status_comment.body
fcp.fk_initiating_comment
initiating_comment.body
or something of the sort Outstanding questions:
IIRC one of the additional problems is that there's no way to ask the rfcbot API for issues which are under FCP for a particular team (though that information is presumably tracked somehow -- it's displayed on rfcbot.rs)
- @Mark-Simulacrum
would you be able to link to wherever that's displayed in rfcbot? Or are you saying it's in the rfcbot.rs/api/all endpoint already?- Looks like you're talking about the rfcbot homepage
- @Mark-Simulacrum & @nikomatsakis — what's the significance of knowing the particular team? Does this affect which FCPs get rendered? (Ties in with the first question about where the lang agenda outputs currently)
we know the set of issues already based on github labels, can we ask rfcbot for the markdown test for a particular issue? that'd be a good start.
- @nikomatsakis what's the relevance of knowing the labels? Is that what determines the relevant team in an FCP?
- @nikomatsakis what do you mean by a "markdown test"?
Oh also... another outstanding question:
fcp.disposition == "merge"
?@chazkiker2
The template actually lives in this repo! It is generated by cargo run --bin lang-agenda
or something like that. The template itself is here but the binary and wrapper is invoked here:
which in turn uses this helper:
what's the significance of knowing the particular team? Does this affect which FCPs get rendered? (Ties in with the first question about where the lang agenda outputs currently)
I actually don't think you need to worry about that. The agenda is already identifying issues tagged with the appropriate team. We just want to add extra info for particular issues.
what's the relevance of knowing the labels? Is that what determines the relevant team in an FCP?
Yes. The issues tagged with T-lang are those that are waiting on a lang team decision.
what do you mean by a "markdown test"?
Typo, I meant "markdown text"
@chazkiker2
Since we're only trying to get proposed FCPs ...
Right now, the list of issues is generated by querying github. I think that is working fine, so potentially we could just consult the RFC bot for particular issues, and not to generate the list itself, but I'm not sure how the API is setup. I also think it would be fine to consult the list -- the only catch is that some of the things in FCP, I realize, may be using the "rustbot second" system as well, so we'd have to account for that.
(We need to bring those two into more harmony.)
@rustbot claim
One of the items in the lang-agenda dumps out the proposed FCPs:
https://github.com/rust-lang/triagebot/blob/bc415ec002845aa0c90a28df5457b666381769f2/templates/lang_agenda.tt#L46-L51
It would be awesome to include registered comments and the like.