:! generic syntax
#676
:! generic syntax
#676
Conversation
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
There was a problem hiding this comment.
Thanks for this. FWIW, I think this summary is really great. This was a complex issue with a lot of discussion. I think this does a great job of pulling together a good overview of the thinking that went into the decision.
The change LGTM, but we should get at least one more thumbs up here before landing.
|
Is this proposal intended to be an RFC? It's currently categorized as draft. |
Yes, it is an RFC. It appears there are two independent ways it can be marked as a draft? @jonmeow |
github-actions
bot
added
the
proposal rfc
|
Kind of, but switching project columns is the only one that really works. Switching project columns should also take care of the rest. |
proposals/p0676.md
Outdated
| #### Square brackets | ||
|
|
||
| Using `[`...`]` consistently for generics creates the opposite problem of the | ||
| brackets being inconsistent with the deduced or explicit distinction. |
There was a problem hiding this comment.
Maybe I'm the only one confused by this, but aren't you using square brackets for the proposal?
There was a problem hiding this comment.
Square brackets are only used to indicate "deduced" in the proposal. Generics are indicated using :!. A generic parameter can be in the deduced list or in the regular (...) explicit parameter list. For now, the only example we have of non-generic deduced parameters is me: Self, but eventually we will probably have more.
There was a problem hiding this comment.
But what was this alternative? Can you give an example of what was considered?
There was a problem hiding this comment.
@chandlerc Can you recall what was actually considered? I could potentially just delete this little bit if it doesn't match what you actually discussed
There was a problem hiding this comment.
Ah, I think I found the example in the discussion notes, I'll add it to the doc.
There was a problem hiding this comment.
I wrote down the example, but it may have fallen under "We never really broke out of the idea that [...] were for deduced parameters" and wasn't actually considered. Perhaps I should include something in the "alternatives not considered" about figuring out which parameters are deduced the same way C++ does?
There was a problem hiding this comment.
The example matches my memory -- this was exactly parallel to <>s and didn't try to get to a consistent story. I think what you have here is right, and the "not considered" thing at the bottom already captures well the space we didn't spend time exploring.
github-actions
bot
added
proposal accepted
| var to: Vector[i64] = CastAVector[i64](from); | ||
| ``` | ||
|
|
||
| One concern is that use of square brackets will be in the same contexts as |
There was a problem hiding this comment.
@zygoloid Could you remind me of your other concern about this option?
There was a problem hiding this comment.
The other concern was that we might want to put generic / template parameters in the (...) list. Motivation for this can be found here: http://open-std.org/JTC1/SC22/WG21/docs/papers/2019/p1045r1.html
This implements decision #565 to use
T:! Typeto declare generic parameters, andtemplate T:! Typefor template parameters.