[PATCH| ArrayRef'ize Sema::FinalizeDeclaratorGroup, Sema::BuildDeclaratorGroup and Sema::ActOnDocumentableDecls

[Rafael Espíndola]

r185931.

On 8 July 2013 18:20, Robert Wilhelm <robert.wilhelm at gmx.net> wrote:
> I have attached patch with removed space svn diffed against current
> trunk.
> Please commit.
>
> Thanks,
> Robert
>
> On Mon, 2013-07-08 at 10:38 -0400, Rafael Espíndola wrote:
>> Sorry I missed this. Just found it while going over old emails:
>>
>> +      BuildDeclaratorGroup( llvm::MutableArrayRef<Decl *>(BeginEndDecls, 2),
>>
>> No space after the (
>>
>> LGTM with that.
>>
>>
>> On 23 May 2013 17:41, Robert Wilhelm <robert.wilhelm at gmx.net> wrote:
>> > Thanks for the review. I have attached new version of the patch.
>> >
>> >
>> > I looked why there were no test failures:
>> > the off-by-one error in Sema::BuildDeclaratorGroup had no negativ effect
>> > (besides small runtime penalty) because you need minimum two decls to be
>> > different for a diagnostic.
>> >
>> > The Group.slice(1) in Sema::ActOnDocumetableDecls is called in testsuite
>> > at typedefs  but was not able to create a doxygen comments where it
>> > makes a difference.
>> >
>> >
>> > On Fri, 2013-05-17 at 10:39 -0700, David Blaikie wrote:
>> >>
>> >>
>> >>
>> >> On Fri, May 17, 2013 at 6:31 AM, Rafael Espíndola
>> >> <rafael.espindola at gmail.com> wrote:
>> >>         On 11 May 2013 15:26, Robert Wilhelm <robert.wilhelm at gmx.net>
>> >>         wrote:
>> >>         > This patch converts three methods to ArrayRef in Sema.
>> >>         > No functionality change.
>> >>         > Passes make test on x86_64-unknown-linux-gnu
>> >>
>> >>
>> >>         -  if (TypeMayContainAuto && NumDecls > 1) {
>> >>         +  if (TypeMayContainAuto && !Group.empty()) {
>> >>
>> >>         Should be 'Group.size() > 1' no?
>> >>
>> >>         -      Group++;
>> >>         -      NumDecls--;
>> >>         +      Group.slice(1);
>> >>
>> >>         slice returns a new array ref, it doesn't modify the current
>> >>         one.
>> >>
>> >>
>> >> & if these mistakes weren't causing test failures - perhaps you could
>> >> figure out which tests are missing & add them :)
>> >>
>> >> - David
>> >
>