Document the new individual const fn feature gates

[durka]

Introduced in #43017. Do they need individual tracking issues as well?

[durka]

Also, this information should be exposed in rustdoc somehow. cc @GuillaumeGomez

[GuillaumeGomez]

I'm not sure how though...

[est31]

@durka features can share tracking issues. The other way doesn't work though, a feature can only have at most one tracking issue.

[durka]

@est31 these aren't normal features so we can do whatever we want :) My plan is to create one tracking issue for all of them, and if one function turns out to need a lot of discussion it can be split off.

[est31]

@durka all I'm saying that there is precedent for features sharing tracking issues, but there is no precedent for features having multiple tracking issues. Why would that make sense in the first place? It would just confuse people to have multiple tracking issues for one feature gate. Anyway, I definitely agree with having one tracking issue for all the const fn feature gates.

[durka]

Right. Maybe it should just stay as #24111 if we can edit the top post.

…

On Sun, Sep 17, 2017 at 2:58 PM, est31 ***@***.***> wrote: @durka <https://github.com/durka> all I'm saying that there is precedent for features sharing tracking issues, but there is no precedent for features having multiple tracking issues. Why would that make sense in the first place? It would just confuse people to have multiple tracking issues for one feature gate. Anyway, I definitely agree with having one tracking issue for all the const fn feature gates. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#44644 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAC3n9jVSG3U8hKWaPqoQUxJWw-yfL3lks5sjWvngaJpZM4PaA86> .

[steveklabnik]

I'm a bit confused; introducing new features should require new pages in the unstable book, it doesn't seem like they exist though?

tagging as medium to sort this out

[est31]

@steveklabnik it autogenerates missing files during doc generation, in a step before the actual md is converted to html so the stub shows up in the rendered output. This makes contributing to the compiler easier because its less red tape to add a new feature. Users will still see the stub, so nothing has changed for them. If autogeneration doesn't work, please ping me!

[steveklabnik]

If autogeneration doesn't work, please ping me!

Yes, I think that's the problem, as we're not seeing pages for those feature gates.

[durka]

They aren't real feature gates, they are the new `#[rustc_const_unstable] ` attribute, so likely they never got added to the tooling/lints.

…

On Oct 31, 2017 16:04, "Steve Klabnik" ***@***.***> wrote: If autogeneration doesn't work, please ping me! Yes, I think that's the problem, as we're not seeing pages for those feature gates. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#44644 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAC3nxs--UwLRXQ0mSIOdj-BkpfQ8qO2ks5sx31rgaJpZM4PaA86> .

[steveklabnik]

#45671 should have made this possible, and we don't require docs for any particular flags, so closing!