Document `label_break_value` in the reference

[jyn514]

r? @ehuss

[nbdd0121]

Shouldn't this be called labelled block expression instead of labelled break expression?

[ciphergoth]

Is this really a loop expression? It doesn't loop.

[Havvy]

I take it that it is documented in loop expressions because it can only be used in loop expression context. Which I think is wrong. I don't see label-break-value as introducing a new kind of expression but rather modifies the semantics of just a block expression and break expressions. If I'm not too tired, I think I might create a competing pull request with how I think it should be documented; explaining it fully would be too much work.

Also, please oh please don't call this a labeled expression if we end up keeping this as its own expression kind. At least call it a labeled block expression or something.

[Havvy]

I, uh, didn't write the PR today. I did think about perspectives some more, and now there's too many perspectives in my head. I wrote about them at https://rust-lang.zulipchat.com/#narrow/stream/237824-t-lang.2Fdoc/topic/Perspectives.20on.20Break.20Label.20Value

[Havvy]

I'm nominating this issue as per the question raised on Zulip.

[nikomatsakis]

My two cents--

I think the approach in this PR seems fine, but I think we should have a link from wherever blocks are documented saying "labeled blocks are described under the loop section".

It's not clear that it's the best approach (if there is such a thing).

[jyn514]

I tried to address @ehuss and @nikomatsakis's concerns. I renamed "label expression" to "label block expression".

[pnkfelix]

@jyn514 It want to echo the comments of others that it feels really weird to document labelled blocks in a section named "Loop expressions"

Could we rename the "Loop expressions" chapter to something that doesn't indirectly imply that a labelled block somehow is a looping construct?

E.g. "Loop expressions and non-local jumps" or "Loop expressions and non-local control flow"? (Or even just "Non-local control flow" or "Complex control flow expressions"?)

[nbdd0121]

Or "breakable expressions"?

[pnkfelix]

I like "Loops and breakable expressions", that rolls off the tongue nicely.

("Loops and labelled block expressions" would also be fine, but I think "breakable expressions" jumps out with the purpose of the concept more immediately.)

I would also be happy with "Loops and other breakable expressions", to try to indicate why these things are in the same section.

[jyn514]

I don't have objections to @pnkfelix's suggestion, but I don't know when I'll have time to apply them. Happy for someone else to push to my branch.

[pnkfelix]

Happy for someone else to push to my branch.

Done!