[2025-06-lwg-21] P3149R11 async_scope - Creating scopes for non-sequential concurrency #8027
Conversation
That's not really a hard conflict due to name lookup rules, right? |
[exec.associate]p12 Renamed "scope_token" to "scope_tkn" due to libconcept of
the same name.
[exec.spawn.future]p2 [exec.spawn]p2 Employ bullets.
[exec.stop.when]p3 [exec.counting.scope]p11 Merged paragraphs via bullets for
clarification ("sndr" and "token" aren't defined otherwise and get confused
with the \exposids of the same name).
[exec.counting.scopes.general] Added missing \expos comments.
[exec.counting.scopes.general] Fixed "close" in scope-state-type to "closed".
[exec.counting.scopes.general]p1.3 Changed "the scope" to "scope".
burblebee
force-pushed
the
motions-2025-06-lwg-21
branch
from
ed9536d to
109df4b
Compare
|
|
|
||
| \item% | ||
| when an object scope of type |
There was a problem hiding this comment.
| when an object scope of type | |
| when an object \tcode{scope} of type |
| \tcode{std::execution::simple_counting_scope} or | ||
| \tcode{std::execution::counting_scope} | ||
| is destroyed and | ||
| \tcode{scope.state} is not equal to |
There was a problem hiding this comment.
| \tcode{scope.state} is not equal to | |
| \tcode{scope.\exposid{state}} is not equal to |
|
|
||
| auto query(execution::get_scheduler_t) const noexcept { | ||
| return @\exposid{loop}@->get_scheduler(); | ||
| \indexlibraryglobal{execution::\exposid{associate-data}}% |
There was a problem hiding this comment.
To my knowledge, we don't index exposition-only declarations.
| \begin{itemdescr} | ||
| \pnum | ||
| \constraints | ||
| copy_constructible<\exposid{wrap-sender}> is \tcode{true}. |
There was a problem hiding this comment.
| copy_constructible<\exposid{wrap-sender}> is \tcode{true}. | |
| \tcode{\libconcept{copy_constructible}<\exposid{wrap-sender}>} is \tcode{true}. |
Or equivalently
\exposid{wrap-sender} satisfies \libconcept{copy_constructible}but it's better to preserve the wording as is here.
| @\exposid{state}@->@\exposid{loop}@.finish(); | ||
| \end{codeblock} | ||
| Initializes \exposid{sndr} with \tcode{std::move(other.\exposid{sndr})} and | ||
| initializes \exposid{token} with \tcode{std::move(other.\linebreak{}\exposid{token})} and |
There was a problem hiding this comment.
| initializes \exposid{token} with \tcode{std::move(other.\linebreak{}\exposid{token})} and | |
| initializes \exposid{token} with \tcode{std::move(other.\brk{}\exposid{token})} and |
\linebreak forces a linebreak, which can cause problems when the paragraph is changed in the future. We have \brk{} for this (see macros.tex) which adds a discretionary line break.
| \tcode{\libconcept{scope_token}<Token>}, or | ||
| \tcode{\exposconcept{queryable}<Env>} | ||
| are not satisfied, | ||
| the expression \tcode{spawn(\linebreak{}sndr, token, env)} is ill-formed. |
There was a problem hiding this comment.
| the expression \tcode{spawn(\linebreak{}sndr, token, env)} is ill-formed. | |
| the expression \tcode{spawn(\brk{}sndr, token, env)} is ill-formed. |
| auto token = std::move(this->@\exposid{token}@); | ||
|
|
||
| @\exposid{destroy}@(); | ||
| token.disassociate(); |
There was a problem hiding this comment.
The paper has a blank line between each statement. I think we should either do what the paper says, or just get rid of all. The blank line after auto token is not really necessary.
|
|
||
| \pnum | ||
| The expression \tcode{spawn(sndr, token)} is expression-equivalent to | ||
| \tcode{spawn(sndr, token, execution::env<>(\linebreak{}))}. |
There was a problem hiding this comment.
| \tcode{spawn(sndr, token, execution::env<>(\linebreak{}))}. | |
| \tcode{spawn(sndr, token, execution::env<>(\brk{}))}. |
| has type \exposid{run-loop-scheduler}, and | ||
| compares equal to the \exposid{run-loop-\newline scheduler} instance | ||
| from which \exposid{sndr} was obtained. | ||
| given an lvalue token of type (possibly const) \tcode{Token}, |
There was a problem hiding this comment.
| given an lvalue token of type (possibly const) \tcode{Token}, | |
| given an lvalue \tcode{token} of type (possibly const) \tcode{Token}, |
| \pnum | ||
| For purposes of determining the existence of a data race, | ||
| \tcode{get_token}, | ||
| \tcode{close}, \tcode{join}, |
There was a problem hiding this comment.
| \tcode{close}, \tcode{join}, | |
| \tcode{close}, | |
| \tcode{join}, |
|
I think I've missed a few |
[exec.associate]p12 Renamed "scope_token" to "scope_tkn" due to libconcept of
the same name.
[exec.spawn.future]p2 [exec.spawn]p2 Employ bullets. [exec.stop.when]p3 [exec.counting.scope]p11 Merged paragraphs via bullets for
clarification ("sndr" and "token" aren't defined otherwise and get confused
with the \exposids of the same name).
[exec.counting.scopes.general] Added missing \expos comments. [exec.counting.scopes.general] Fixed "close" in scope-state-type to "closed". [exec.counting.scopes.general]p1.3 Changed "the scope" to "scope".
Fixes #7958.
Fixes cplusplus/papers#1800