mclow.lists closed this revision.
mclow.lists marked an inline comment as done.
mclow.lists added a comment.
Committed as revision 298681.
https://reviews.llvm.org/D31163
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/
EricWF accepted this revision.
EricWF added a comment.
This revision is now accepted and ready to land.
LGTM. I'll send something to the reflectors about the explicit deduction guides
over the weekend.
https://reviews.llvm.org/D31163
___
cfe-commit
mclow.lists marked 8 inline comments as done.
mclow.lists added inline comments.
Comment at: include/mutex:176
+template unique_lock(unique_lock)
+-> unique_lock; // C++17
mclow.lists wrote:
> EricWF wrote:
> > This should be guarded behind a feature test m
mclow.lists updated this revision to Diff 92896.
mclow.lists added a comment.
Removed the deduction guides. Guarded the tests for the deduction guides with
#ifdefs.
https://reviews.llvm.org/D31163
Files:
include/__config
include/__mutex_base
include/mutex
include/shared_mutex
test/l
EricWF added inline comments.
Comment at: include/mutex:177
+template unique_lock(unique_lock)
+-> unique_lock; // C++17
+
mclow.lists wrote:
> And what's with the 'M'? (here and below)
No idea, but I would remove the explicit deduction guides from this patc
mclow.lists added inline comments.
Comment at: include/mutex:176
+template unique_lock(unique_lock)
+-> unique_lock; // C++17
EricWF wrote:
> This should be guarded behind a feature test macro. I would suggest adding
> this to `__config`.
>
> ```
> #if !d
EricWF added a comment.
So Clang only very recently added support for deduction guides. We'll need to
add `_LIBCPP_HAS_NO_DEDUCTION_GUIDES` and use it to guard the declarations and
the tests.
Also I think the explicit deduction guides are incorrect and unneeded. The
deduction guides you declar
mclow.lists created this revision.
WG21 has decided that the ABI-breaking `lock_guard` proposal that we (@EricWF)
implemented (and hid behind an ABI-break macro) is not the way to go. Instead,
we're introducing a new class, `scoped_lock`, which is variadic, and leaving
`lock_guard` as non-varia
