[llvm] r289579 - ADT: Add OwningArrayRef class.

John McCall via llvm-commits llvm-commits at lists.llvm.org
Mon Dec 19 21:57:51 PST 2016


> On Dec 19, 2016, at 9:47 PM, David Blaikie <dblaikie at gmail.com> wrote:
> On Mon, Dec 19, 2016 at 9:24 PM Peter Collingbourne <peter at pcc.me.uk <mailto:peter at pcc.me.uk>> wrote:
> Sure, I understood what you meant. I meant that I wouldn't take a position on whether avoiding the cost of the capacity and the reserve area is worth it.
> 
> (If pressed I think I'd say no, the average tu doesn't have that many vtables, and there are far more egregious wastes of memory in llvm anyway (e.g. llvm::DIE) that we should be concentrating on first, but while I was adding another thing to vtables I figured it wouldn't hurt to be consistent with the others, then rule of three kicked in so seemed reasonable to add the abstraction.)
> 
> Yeah, that's pretty much how I feel too.
> 
> Richard, Chandler - I seem to recall this has come up before (whether or not LLVM would benefit from a dynarray like abstraction) & I don't remember the backstory on the standards committee for dynarary (which I would've only heard second hand from one of you, I think). Any extra context/thoughts you could share here, briefly?

Is your argument really that we should intentionally pessimize something because the committee decided not to standardize a similar container?  LLVM's entire ADT library is basically a laundry list of micro-optimizations that we felt at some point or another had advantages over what the STL provides.

John.

> 
> - Dave
>  
> 
> Peter
> 
> 
> On Dec 19, 2016 20:29, "David Blaikie" <dblaikie at gmail.com <mailto:dblaikie at gmail.com>> wrote:
> What I mean is: compare OwningArrayRef to std::vector, not OwningArrayRef to manual memory management
> 
> On Mon, Dec 19, 2016 at 8:07 PM Peter Collingbourne <peter at pcc.me.uk <mailto:peter at pcc.me.uk>> wrote:
> I don't really have a strong opinion about it. It wraps up some manual memory management code we used to have in the vtable builder, although I couldn't say how beneficial that memory management really is.
> 
> Peter
> 
> On Mon, Dec 19, 2016 at 7:56 PM, David Blaikie <dblaikie at gmail.com <mailto:dblaikie at gmail.com>> wrote:
> Is this really worth having compared to std::vector? std::dynarray was rejected from standardization for that reason, if I understand/heard correctly (& OwningArrayRef seems similar to std::dynarray)
> 
> On Tue, Dec 13, 2016 at 12:34 PM Peter Collingbourne via llvm-commits <llvm-commits at lists.llvm.org <mailto:llvm-commits at lists.llvm.org>> wrote:
> Author: pcc
> Date: Tue Dec 13 14:24:24 2016
> New Revision: 289579
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=289579&view=rev <http://llvm.org/viewvc/llvm-project?rev=289579&view=rev>
> Log:
> ADT: Add OwningArrayRef class.
> 
> This is a MutableArrayRef that owns its array.
> I plan to use this in D22296.
> 
> Differential Revision: https://reviews.llvm.org/D27723 <https://reviews.llvm.org/D27723>
> 
> Modified:
>     llvm/trunk/include/llvm/ADT/ArrayRef.h
> 
> Modified: llvm/trunk/include/llvm/ADT/ArrayRef.h
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/ArrayRef.h?rev=289579&r1=289578&r2=289579&view=diff <http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/ArrayRef.h?rev=289579&r1=289578&r2=289579&view=diff>
> ==============================================================================
> --- llvm/trunk/include/llvm/ADT/ArrayRef.h (original)
> +++ llvm/trunk/include/llvm/ADT/ArrayRef.h Tue Dec 13 14:24:24 2016
> @@ -413,6 +413,25 @@ namespace llvm {
>      }
>    };
> 
> +  /// This is a MutableArrayRef that owns its array.
> +  template <typename T> class OwningArrayRef : public MutableArrayRef<T> {
> +  public:
> +    OwningArrayRef() {}
> +    OwningArrayRef(size_t Size) : MutableArrayRef<T>(new T[Size], Size) {}
> +    OwningArrayRef(ArrayRef<T> Data)
> +        : MutableArrayRef<T>(new T[Data.size()], Data.size()) {
> +      std::copy(Data.begin(), Data.end(), this->begin());
> +    }
> +    OwningArrayRef(OwningArrayRef &&Other) { *this = Other; }
> +    OwningArrayRef &operator=(OwningArrayRef &&Other) {
> +      delete this->data();
> +      this->MutableArrayRef<T>::operator=(Other);
> +      Other.MutableArrayRef<T>::operator=(MutableArrayRef<T>());
> +      return *this;
> +    }
> +    ~OwningArrayRef() { delete this->data(); }
> +  };
> +
>    /// @name ArrayRef Convenience constructors
>    /// @{
> 
> 
> 
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at lists.llvm.org <mailto:llvm-commits at lists.llvm.org>
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits <http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits>
> 
> 
> 
> -- 
> -- 
> Peter
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20161219/4d657f11/attachment.html>


More information about the llvm-commits mailing list