- User Since
- Jun 26 2014, 12:44 PM (269 w, 2 d)
Thu, Aug 22
Sorry for being late to the party, I've looked at the codege of this and other approachs. what we want is a single global than initializes all of objects we need. Right now we have function local static then may may not have make guards and that's less than ideal.
Wed, Aug 21
I don't fully understand this patch. The point of always declaring the interface was to enforce that it matches exactly what we expect. I'll grant that probably wasn't working, but can you explain more about?
Tue, Aug 20
Mon, Aug 12
Fri, Aug 9
Thu, Aug 8
Sun, Aug 4
Sat, Aug 3
wow. This is a fantastic amount of work. Thank you. No longer will libc++ accidentally break the pretty printers!
Fri, Aug 2
Or alternatively, there's an example of how to fix this in memory_resource.cpp https://github.com/llvm-mirror/libcxx/blob/master/src/experimental/memory_resource.cpp#L67-L96
I think a better fix would be to apply [[clang::no_destroy]] to the category variable definitions instead.
Thu, Aug 1
Wed, Jul 31
Oh, I forget to mention:
I still have my concerns about how we find, package, and install the PSTL. But after offline discussions with Louis, it's clear this is on his radar as well, and improvements will be made going forward.
This is a good first step.
Tue, Jul 30
LGTM after moving the define to <__config>.
Mon, Jul 29
I have some concerns I would like to see addressed before this lands again.
Committed in r367263. Thanks for the change, sorry about the delay.
Sun, Jul 28
Sat, Jul 27
Jul 18 2019
@ldionne pointed out to me that I'm all kinds of wrong.
ARM is the correct monacer for this configuration. The spec we're implementing was written by ARM.
This change isn't strictly correct. A lot of these constructors were externally instantiated before we made them templates.
I suspect this will break the ABI.
Jul 17 2019
Thanks for bringing this up. Indeed this change really hurts TSAN.
I'll look for ways to work around this issue.
Jul 16 2019
For sanitizers I wonder if we could continue to catch use after destrot without losing triviality. Doesn't MSAN poison the objects memory at the end of the destructor?