v8config.h: set V8_CC_MSVC when building with clang-cl

v8config.h: introduce V8_LIBC_MSVCRT

Use this for detecting MSVCRT library features instead of
V8_CC_MSVC.

One use case for this is when compiling with Clang together with the
MSVC library. In that case, V8_CC_MSVC will be false, but V8_LIBC_MSVCRT
will be true.

BUG=82385
LOG=n
R=bmeurer@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=18888

[hans, 2014-01-23 17:30:53]

Please take a look.

(Also, I don't have commit access, so someone will need to land this for me.)

[Benedikt Meurer, 2014-01-24 09:17:45]

Upload is broken, please repeat upload.

[hans, 2014-01-24 16:44:02]

On 2014/01/24 09:17:45, Benedikt Meurer wrote:
> Upload is broken, please repeat upload.

Sorry about that, not sure what happened.

I've uploaded again.

[Benedikt Meurer, 2014-01-27 06:25:37]

LGTM with nit.

https://codereview.chromium.org/145593003/diff/80001/include/v8config.h
File include/v8config.h (right):

https://codereview.chromium.org/145593003/diff/80001/include/v8config.h#newco...
include/v8config.h:237: # if defined(_MSC_VER)
Please add a comment why this is necessary.

[hans, 2014-01-27 17:23:45]

On 2014/01/27 06:25:37, Benedikt Meurer wrote:
> LGTM with nit.
Thanks! Can you commit this for me?

> https://codereview.chromium.org/145593003/diff/80001/include/v8config.h
> File include/v8config.h (right):
> 
>
https://codereview.chromium.org/145593003/diff/80001/include/v8config.h#newco...
> include/v8config.h:237: # if defined(_MSC_VER)
> Please add a comment why this is necessary.
Done.

[Benedikt Meurer, 2014-01-28 05:34:03]

Hm, thinking about this again: The problem is that we depend on V8_CC_MSVC to
actually detect the MS C runtime, right? Wouldn't it make sense to introduce
V8_LIBC_MSVCRT and fix relevant uses of V8_CC_MSVC?

[Sven Panne, 2014-01-28 07:42:07]

On 2014/01/28 05:34:03, Benedikt Meurer wrote:
> Hm, thinking about this again: The problem is that we depend on V8_CC_MSVC to
> actually detect the MS C runtime, right? Wouldn't it make sense to introduce
> V8_LIBC_MSVCRT and fix relevant uses of V8_CC_MSVC?

+1 to that. A quick look into our source code reveals that all but one use of
V8_CC_MSVC actually check for library features, not compiler features. The only
real compiler check is for integral suffixes.

Cleaning this up would bring our Windows support in line with the usual
distinction on all other platforms (like GCC vs. clang, POSIX vs. BSD, etc.).

[hans, 2014-01-28 18:58:02]

On 2014/01/28 07:42:07, Sven Panne wrote:
> On 2014/01/28 05:34:03, Benedikt Meurer wrote:
> > Hm, thinking about this again: The problem is that we depend on V8_CC_MSVC
to
> > actually detect the MS C runtime, right? Wouldn't it make sense to introduce
> > V8_LIBC_MSVCRT and fix relevant uses of V8_CC_MSVC?
> 
> +1 to that. A quick look into our source code reveals that all but one use of
> V8_CC_MSVC actually check for library features, not compiler features. The
only
> real compiler check is for integral suffixes.
> 
> Cleaning this up would bring our Windows support in line with the usual
> distinction on all other platforms (like GCC vs. clang, POSIX vs. BSD, etc.).

Sounds reasonable. I've uploaded a new patch and edited the description.

Please take another look.

[Benedikt Meurer, 2014-01-28 19:09:56]

LGTM, thanks. I'll land it.

[Benedikt Meurer, 2014-01-28 19:11:23]

Committed patchset #4 manually as r18888 (presubmit successful).