[llvm-commits] [PATCH] Add APInt(numBits, ArrayRef<uint64_t> bigVal) constructor (issue4749046)

[Eli Friedman]

On Mon, Jul 18, 2011 at 9:57 AM, Jeffrey Yasskin <jyasskin at google.com> wrote:
> Here's the patch.
>
> On Sat, Jul 16, 2011 at 10:51 PM,  <jyasskin at gmail.com> wrote:
>> Reviewers: eli.friedman_gmail.com,
>>
>> Message:
>> Initial diff at
>> http://codereview.appspot.com/download/issue4749046_1.diff
>>
>> Description:
>> Add APInt(numBits, ArrayRef<uint64_t> bigVal) constructor to prevent
>> future ambiguity errors like the one corrected by r135261.  Migrate all
>> LLVM callers of the old constructor to the new one.
>>
>>
>> Please review this at http://codereview.appspot.com/4749046/
>>
>> Affected files:
>>   M include/llvm/ADT/APInt.h
>>   M lib/AsmParser/LLLexer.cpp
>>   M lib/Bitcode/Reader/BitcodeReader.cpp
>>   M lib/CodeGen/SelectionDAG/FastISel.cpp
>>   M lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp
>>   M lib/CodeGen/SelectionDAG/SelectionDAG.cpp
>>   M lib/ExecutionEngine/ExecutionEngine.cpp
>>   M lib/Support/APFloat.cpp
>>   M lib/Support/APInt.cpp
>>   M lib/VMCore/ConstantFold.cpp
>>   M lib/VMCore/Core.cpp
>>   M unittests/ADT/APIntTest.cpp

@@ -342,7 +353,8 @@ public:

     if (isSingleWord())
       return isUIntN(N, VAL);
-    return APInt(N, getNumWords(), pVal).zext(getBitWidth()) == (*this);
+    return APInt(N, ArrayRef<uint64_t>(pVal,
getNumWords())).zext(getBitWidth())
+      == (*this);
   }

   /// @brief Check if this APInt has an N-bits signed integer value.

makeArrayRef just landed; use it. :)  Same in a couple other places.

Otherwise, looks fine.

As a followup, could you eliminate both the "APInt(unsigned numBits,
unsigned numWords, const uint64_t bigVal[])" and the "APInt(unsigned
numBits, uint64_t val, bool isSigned = false)" constructors?  (It
should just require ArrayRef-izing and adding an
"APInt::getSigned(unsigned numBits, uint64_t val)" helper.)

-Eli