[PATCH] D21563: [PM] refactoring of LoopAccessInfo /NFC

[Xinliang David Li via llvm-commits]

Ok -- I plan to explicitly define move ctors in the second stage change.

thanks,

David

On Wed, Jun 22, 2016 at 3:48 PM, Sean Silva <chisophugis at gmail.com> wrote:

> silvas added a subscriber: silvas.
>
> ================
> Comment at: include/llvm/Analysis/LoopAccessAnalysis.h:617-623
> @@ -616,9 +616,9 @@
>
>    /// We need to check that all of the pointers in this list are disjoint
>    /// at runtime.
> -  RuntimePointerChecking PtrRtChecking;
> +  std::unique_ptr<RuntimePointerChecking> PtrRtChecking;
>
>    /// \brief the Memory Dependence Checker which can determine the
>    /// loop-independent and loop-carried dependences between memory
> accesses.
> -  MemoryDepChecker DepChecker;
> +  std::unique_ptr<MemoryDepChecker> DepChecker;
>
> ----------------
> anemet wrote:
> > Please add a comment saying that these are pointers to facilitate the
> move ctor.
> This might not be relevant, but FYI there is a gotcha with MSVC 2013 where
> it can't synthesize a move ctor. Just a heads up in case there are bot
> problems with this patch.
>
> This caused me problems when porting JumpThreading. See
> http://reviews.llvm.org/rL272607 in particular the manually defined move
> constructor with the comment "// Hack for MSVC 2013 which seems like it
> can't synthesize this.".
>
> The fix there is based on the earlier r216244 that gave insight into the
> problem.
>
>
> http://reviews.llvm.org/D21563
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160622/3a89354c/attachment.html>