- User Since
- Feb 11 2015, 3:26 PM (331 w, 23 h)
LGTM, but just out of curiosity, will this stay unsupported forever, or will this be implemented eventually?
Associated release blocking bug: http://llvm.org/PR50734
Associated release blocker bug: http://llvm.org/PR50758
Associated release blocker bug: http://llvm.org/PR50757
Just to record this publicly:
@phosek Could you please apply this change locally and run whatever you ran that caused you to revert last week? I think we should have fixed our stuff, but I really want to make sure we're not going to break you again, because it's extremely painful for us to revert and re-apply such a large change (and it messes up the history pretty badly, too).
Update to https://reviews.llvm.org/D104171.
Thanks, will ship once CI is green. I'll also cherry-pick this one to release/12.x.
(and CI passing once you rebase)
LGTM with my comments applied.
LGTM if CI passes!
@tstellar Permission to merge this into release/12.x once the CI is passing.
commit e7dac564cd0ed9dee74ef972c46622743d90915d (HEAD -> release/12.x, origin/release/12.x) Author: Louis Dionne <firstname.lastname@example.org> Date: Tue Jun 15 17:55:27 2021 -0400
The remaining CI issues already exist before this patch. I'll try fixing them subsequently, but this is good to go.
Tue, Jun 15
Fix CI issues.
Fix CI failures
I created another take on this here: https://reviews.llvm.org/D104328.
This is a manual cherry-pick because the codebase has diverged quite a bit.
Address review comments and clarify intent about supported versions.
Update XFAIL for libc++abi on recent GCC
Mon, Jun 14
Thanks a lot! Indeed, we don't have access to FileCheck in the libc++ tests. I had once looked into doing that but I ran into issues and we didn't have a clear use case so I decided not to spend more time on it. Looks like this will do for the time being.
Good catch on the end.cpp rename - we were not running that test.
LGTM but can you please move the operator& changes to the other patch? Also, I think this does reduce the usefulness of __function_like by quite a bit, cause now we have to think about adding a declaration in the function objects *and* derive from __function_like. I think we should fix the underlying modules bug (seriously, modules shouldn't have anything to do with this!) and more operator& back into __function_like.
Thanks for the cleanup.
Include changes LGTM, but I don't understand the change to __function_like.
Did we have any uses of test_hash with a hash that's not std::hash?
Fri, Jun 11
- CI needs to pass
- Please put the GCC-next CI fixes in a different patch, and don't touch the iterator.ops failures (I'm doing that in D103272).
- We normally add .verify.cpp tests to check that we've applied the deprecated attribute properly to stuff -- here this should apply to the typedefs. It's kinda nitpicky but I'd like it if you could add them.
Address review comments and tweak tests for advance too.
Pending passing CI, of course.
I think it's necessary to have the ABI macro, however I really think we should stop adding those _LIBCPP_ENABLE_CXX20_REMOVED_XYZ macros. If someone is opting into using C++20, then they are opting into these removals as well. At that point fixing their code is probably not much harder than flipping that macro.
Thu, Jun 10
Ship-it. I'll be looking forward to a patch that makes insert_iterator up-to-spec for C++20.
I missed that un-applied comment in my LGTM a second ago. Still LGTM, but why not fix the FIXME now?
Thanks! Now we'll be able to update all the in-progress views to this paper.
For some reason this went entirely under my radar.
Rebase onto main with the new triple-based availability.
Per offline discussion,