[lld] r228968 - [ELF] Insert wrap symbols into a set.

[Chandler Carruth]

On Tue, Feb 17, 2015 at 2:24 PM, Rui Ueyama <ruiu at google.com> wrote:

> On Fri, Feb 13, 2015 at 11:55 AM, Chandler Carruth <chandlerc at google.com>
> wrote:
>
>>
>> On Fri, Feb 13, 2015 at 11:30 AM, Rui Ueyama <ruiu at google.com> wrote:
>>
>>> Returning a const reference to a set is not wrong at all, and it
>>> involves less moving parts, no?
>>>
>>> The other reason why I prefer a set over range here is because I don't
>>> want to see more uses of range in LLD at this moment. I believe range is
>>> designed after N3350, which didn't make it to the standard yet. It doesn't
>>> feel great that this generic library lives under LLD. If we really need
>>> this one, we should move the file to LLVM support directory. Otherwise, I
>>> don't like to build a local rule to use range in LLD. After all other LLVM
>>> projects are not using this range.
>>>
>>
>> FWIW, LLVM projects are using ranges pretty heavily now, and we added
>> llvm::iterator_range to LLVM's ADT library. So I'm pretty happy with
>> increased usage of ranges in LLD.
>>
>
> Thanks, Chandler. Maybe we should replace all uses of lld::range with
> llvm::iterator_range? Looks like there's no reason to not do.
>

I'm generally happy with the direction of sinking this generic
functionality out of LLD and into LLVM. At the time it went in, LLD was
using C++11 in places that LLVM wasn't (IIRC) and that was the motivation
for keeping it in LLD. Today, that doesn't make sense.

I've not looked at LLD's generic utilities here so i can't comment on the
best way to do this or what they'll end up looking like, but I'm vaguely in
favor of the direction.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20150217/474e2270/attachment.html>