No, unique address! - #1363
Conversation
MSVC is finally getting around to implementing `[[no_unique_address]]`, which will result in class layout differing with Clang. Let's comment out the uses that are not yet so commented until we reach compiler parity.
Stephan T. Lavavej (StephanTLavavej)
left a comment
There was a problem hiding this comment.
Looks good to me. I checked out the branch and searched to verify that all occurrences of no_unique_address are commented out.
Michael Schellenberger Costa (miscco)
left a comment
There was a problem hiding this comment.
Looks good, am I understanding it correctly that the feature is WIP in the MSVC side and we do not want to interfere with their work?!
|
My understanding is to avoid ABI/ODR issues between TUs compiled with Clang and TUs compiled with MSVC, until both reach parity. |
|
Charles Milette (@sylveon) Yep. I've filed #1364 to track that investigation (anyone can help by determining whether Clang supports |
|
Thanks for addressing this unique problem no-more. (There's no functional change here, so not Changelogging.) |
MSVC is finally getting around to implementing
[[no_unique_address]], which will result in class layout differing with Clang. Let's comment out the uses that are not yet so commented until we reach compiler parity.Related to MSVC-PR-280491, which implements
[[no_unique_address]]"for real" in MSVC.