[cfe-commits] [PATCH] AST matchers

[Douglas Gregor]

On Jun 29, 2012, at 2:33 PM, David Blaikie <dblaikie at gmail.com> wrote:

> On Fri, Jun 29, 2012 at 2:30 PM, Douglas Gregor <dgregor at apple.com> wrote:
>> 
>> On Jun 29, 2012, at 2:29 PM, David Blaikie <dblaikie at gmail.com> wrote:
>> 
>>> On Fri, Jun 29, 2012 at 2:05 PM, Manuel Klimek <klimek at google.com> wrote:
>>>> On Fri, Jun 29, 2012 at 10:40 PM, Douglas Gregor <dgregor at apple.com> wrote:
>>>>> 
>>>>> CamelCase is fine for these. I notice that there's some inconsistency with
>>>>> 'has' names, e.g., hasDescendant vs. HasType.
>>>> 
>>>> 
>>>> Update: Chandler voted for using llvm coding style (and I agree). The
>>>> argument is that the callable classes we have are just classes to work
>>>> around missing variadic templates and that they will become functions anyway
>>>> once we get C++11 into llvm (in a future, far, far away).
>>> 
>>> Actually - this is a layer on top of Clang the compiler, so why
>>> wouldn't it just use C++11? It doesn't have the same constraints of
>>> buildability with host compilers as the compiler does - you can always
>>> just compile this with Clang.
>> 
>> 
>> Our support for generating code on Windows is still too poor to make this something we depend on.
>> 
>> Unless someone has a Tool that maps variadic templates down to C++03 ;)
> 
> Ah, indeed. Windows. So we'd be limited to MSVC's support (as lld is)
> which wouldn't solve this specific issue, unfortunately. Still,
> potentially makes all this tooling, matcher, and C++11 migration stuff
> able to use the MSVC C++11 features (yay rvalue references at least).


Frankly, I'd rather just wait a bit longer (6 months or a year), then allow all of LLVM to migrate to some reasonably-portable subset of C++11.

	- Doug