Rephrase "possibly const value" in terms of types #1563
Conversation
|
@zygoloid: Should/can we adopt this change? Seems like a useful improvement. |
|
I'm definitely happy to merge all but the first change in containers.tex. For that one change, I'm a little concerned that we're papering over a (pre-existing) wording bug. |
|
@jwakely reports that Marshall Clow is opposed to changing "const value of type X". Editorial meeting consensus: This is technically wrong, though. Interact with LWG. |
|
Sorry for the accidental close. |
zygoloid
force-pushed
the
master
branch
2 times, most recently
from
e3dbfe2 to
1a21a65
Compare
zygoloid
added
the
needs rebase
jwakely
force-pushed
the
possibly-const-value
branch
2 times, most recently
from
3de1933 to
c69af47
Compare
jwakely
removed
the
needs rebase
|
I didn't think the Tentatively Ready poll for 3028 passed, but I've just done recount and it has five votes. But that won't be resolved until the next plenary now, sorry. |
|
Oh OK, but in that case we probably won't need this PR, either, since it'll come back in a paper? But maybe we can reuse the edits here. |
|
I don't think there's going to be a paper making these changes. I'll rebase it to remove the part that conflicts with 3028, so we can make the non-conflicting editorial changes. |
|
Oh, I see, thanks! Let me know when it's ready. |
jwakely
force-pushed
the
possibly-const-value
branch
from
c69af47 to
3d8c66b
Compare
|
So is this editorial now? I'll remove the "lwg" label. |
|
Yes, I think this is purely editorial, and as LWG 3028 shows, LWG is favouring putting "const" on the type, not the value, which is the subject of #1497 and this pull request. I've rebased and removed the conflicts with LWG 3028. That means this won't entirely fix #1497, because the Tentatively Ready resolution for 3028 includes: - - `r` denotes a non-const value of type `X`, and
+ - `s` and `t` denote non-const lvalues of type `X`, and
- `rv` denotes a non-const rvalue of type `X`.i.e. it changes one instance of "non-const value" to "non-const lvalues" and doesn't touch an instance of "non-const rvalue". Those should be fixed after 3028 is applied to the draft, instead of doing it as part of this pull request, to avoid invalidating the 3028 resolution that should get approved at the next plenary. Does that sound reasonable? |
|
Sounds good! We can keep this PR open after merging as a tracking issue. |
|
@jensmaurer, @zygoloid: would you care to provide a corish opinion and assent? |
|
"value of type X or const X" is certainly in-your-face obvious and looks good to me. off-topic: I wish we had some shorthand along the lines of "value of type cv X" or so. |
Fixes #1497