"possibly const" vs "possibly \tcode{const}" vs "possibly \keyword{const}"

[sdutoit]

"possibly const" appears five times in the current draft. "possibly \tcode{const}" appears ten times. Clean this up, make a rule for it.

Update added by @burblebee on 11/15/22:
We've since added several more of the above variations as well as the addition of "possibly \keyword{const}".

[jensmaurer]

Editorial meeting consensus: Similar to inline / public. Thus, we should generally not use "possibly \tcode{const}".

[zygoloid]

See also issue #106.

[jensmaurer]

Postponing until #1497 is applied to avoid editing conflicts.

[burblebee]

Editorial meeting consensus: Similar to inline / public. Thus, we should generally not use "possibly \tcode{const}".

To clarify, all occurrences should become possibly const, correct?

[JohelEGP]

See #4405 (comment) with regards to \keyword. @tkoeppe, can \keyword be rationalized on the wiki?

[burblebee]

I don't think we should use \keyword in this case, since this is basically equivalent to "possibly cv-qualified" (and other variations of "cv"), and we don't include references to those in the index under const. So if we want to recognize "const" here as a keyword (e.g. so we could make them a different color sometime in the future as mentioned in #4405), maybe we need something like a \keywordni (ni for no index)? What about uses of "non-const"?

Adding the "decision required" label back - I'm happy to take this one on if I know what we want to do here.

[tkoeppe]

I think the original decision to say "possibly const" (without markup) is uncontested; shall we just execute that?

[sdutoit]

I think the original decision to say "possibly const" (without markup) is uncontested; shall we just execute that?

Yes please, after these 9+ years! :)

[tkoeppe]

@sdutoit Why did you leave? :-)

[tkoeppe]

@burblebee What do you think needs a decision here? Don't we already have several decisions above?

[burblebee]

I think the original decision to say "possibly const" (without markup) is uncontested; shall we just execute that?

Sure! Sorry this resolution wasn't clear to me, but
"we should generally not use possibly \tcode{const}."
doesn't say what we should do, so thank you for the clarification! There is now the added complexity with \keyword{const}, but I feel the "just say possibly const" resolution works there to.

[jwakely]

Arguably we should just say "type X or const X" instead of "type (possibly const) X".

Not using keyword markup for "possibly const" does look a bit odd when we have:

an lvalue of type (possibly const) X or an rvalue of type const X,

Based on the decision above, one "const" is is code font and one isn't.

[zygoloid]

I think the original decision to say "possibly const" (without markup) is uncontested; shall we just execute that?

Yes please, after these 9+ years! :)

I also favor this direction. Hm, I wonder if we can get Pete Becker's signoff too? :)

[tkoeppe]

@jwakely That one case you pointed out could indeed be rewritten a little to avoid this awkwardness.