Import C++11 headers for unordered_(map|set) when using libc++ in protobuf.

Import C++11 headers for unordered map and set when using libc++ in protobuf's config header to avoid warnings.
    
BUG=366218
TBR=brettw

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=269646

[jfroy, 2014-04-23 19:24:45]

First patch to Chromium, please be nice.

Submitting a set of patches to improve and fix compiling a subset of chromium
for including into the Google Cast SDK for iOS (to support video and audio
casting).

[rmcilroy, 2014-04-24 10:13:31]

I'm not sure on the reasoning behind this change (libc++ has <ext/hash_map> and
<ext/hash_set> AFAIK) - is there a reason you need to use unordered_map/set
instead?  

In any case, I think it would make more sense to make your changes upstream at
https://code.google.com/p/protobuf/ changing config.ac as necessary, otherwise
your changes will be replaced the next time someone runs configure and uploads
the auto-generated config.h.

https://codereview.chromium.org/246633012/diff/1/third_party/protobuf/README....
File third_party/protobuf/README.chromium (right):

https://codereview.chromium.org/246633012/diff/1/third_party/protobuf/README....
third_party/protobuf/README.chromium:19: The config file has been patched to
support both libstdc++ and libc++.
You don't need this line, since you are changing a local file, not one taken
from upstream (see list above).

[jfroy, 2014-04-24 16:15:39]

On 2014/04/24 10:13:31, rmcilroy wrote:
> I'm not sure on the reasoning behind this change (libc++ has <ext/hash_map>
and
> <ext/hash_set> AFAIK) - is there a reason you need to use unordered_map/set
> instead?  

It does, but it warns. Just trying to cleanup.

> 
> In any case, I think it would make more sense to make your changes upstream at
> https://code.google.com/p/protobuf/ changing config.ac as necessary, otherwise
> your changes will be replaced the next time someone runs configure and uploads
> the auto-generated config.h.

protobuf has already been updated and generates a sensible config.h when
targeting
libc++. The issue here is that chromium needs a "adaptive" config.h header that
will
work with the various platforms and configurations that get used to build it.
Hence the
conditional code. Note that the existing STLPORT conditions are not in upstream
protobuf either.

> 
>
https://codereview.chromium.org/246633012/diff/1/third_party/protobuf/README....
> File third_party/protobuf/README.chromium (right):
> 
>
https://codereview.chromium.org/246633012/diff/1/third_party/protobuf/README....
> third_party/protobuf/README.chromium:19: The config file has been patched to
> support both libstdc++ and libc++.
> You don't need this line, since you are changing a local file, not one taken
> from upstream (see list above).

I just thought I should make a note of the change. I can remove that chunk.

[rmcilroy, 2014-04-25 10:42:24]

I see - lgtm modulo the comments below.  I'm not an OWNER though so you will
need an LGTM from an OWNER.

https://codereview.chromium.org/246633012/diff/40001/third_party/protobuf/con...
File third_party/protobuf/config.h (right):

https://codereview.chromium.org/246633012/diff/40001/third_party/protobuf/con...
third_party/protobuf/config.h:2: /* config.h.in.  Generated from configure.ac by
autoheader.  */
Please replace this comment to make it clear that this file was originally
autogenerated, but has since been modified to enable adaptive support for
stlport and libc++.

https://codereview.chromium.org/246633012/diff/40001/third_party/protobuf/con...
third_party/protobuf/config.h:6: /* the name of <hash_map> */
nit - <unordered_map> or <hash_map>

https://codereview.chromium.org/246633012/diff/40001/third_party/protobuf/con...
third_party/protobuf/config.h:29: /* the name of <hash_set> */
nit - <unordered_set> or <hash_set>

[jfroy, 2014-04-25 21:39:23]

I added a note, but the other comments are exactly as found in the generated
config.h from a protobuf checkout and configure, so I'm leaving them alone.

[brettw, 2014-04-26 15:55:12]

https://codereview.chromium.org/246633012/diff/1/third_party/protobuf/config.h
File third_party/protobuf/config.h (right):

https://codereview.chromium.org/246633012/diff/1/third_party/protobuf/config....
third_party/protobuf/config.h:4: #include <ciso646>
This seems a bit mysterious. I assume you picked this to be the simplest
"standard" header so _LIBCPP_VERSION is defined? I think we should document
this.

[jfroy1, 2014-04-26 17:47:50]

Sure, I'll add a comment.

The header file “ciso646″ is required by both the C++03 and C++11
standards, and defined to do nothing.

http://cplusplusmusings.wordpress.com/2013/02/26/c-and-xcode-4-6/
On Sat, Apr 26, 2014 at 8:55, <brettw@chromium.org> wrote:

>
> https://codereview.chromium.org/246633012/diff/1/third_
> party/protobuf/config.h
> File third_party/protobuf/config.h (right):
>
> https://codereview.chromium.org/246633012/diff/1/third_
> party/protobuf/config.h#newcode4
>
third_party/protobuf/config.h:4<https://codereview.chromium.org/246633012/diff/1/third_party/protobuf/config.h#newcode4third_party/protobuf/config.h:4>:
> #include <ciso646>
> This seems a bit mysterious. I assume you picked this to be the simplest
> "standard" header so _LIBCPP_VERSION is defined? I think we should
> document this.
>
> https://codereview.chromium.org/246633012/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[jfroy1, 2014-04-26 17:49:14]

Although
https://qt.gitorious.org/qt/junfs-qt/commit/9deca2ef28e1f593d3882ba0c7e9dd7b0...
we should use <new> instead for MSVS. I assume we care about that
toolchain.
On Sat, Apr 26, 2014 at 10:47, Jean-François Roy <jfroy@google.com> wrote:

> Sure, I'll add a comment.
>
> The header file “ciso646″ is required by both the C++03 and C++11
> standards, and defined to do nothing.
>
> http://cplusplusmusings.wordpress.com/2013/02/26/c-and-xcode-4-6/
> On Sat, Apr 26, 2014 at 8:55, <brettw@chromium.org> wrote:
>
>>
>> https://codereview.chromium.org/246633012/diff/1/third_party
>> /protobuf/config.h
>> File third_party/protobuf/config.h (right):
>>
>> https://codereview.chromium.org/246633012/diff/1/third_party
>> /protobuf/config.h#newcode4
>>
third_party/protobuf/config.h:4<https://codereview.chromium.org/246633012/diff/1/third_party/protobuf/config.h#newcode4third_party/protobuf/config.h:4>:
>> #include <ciso646>
>> This seems a bit mysterious. I assume you picked this to be the simplest
>> "standard" header so _LIBCPP_VERSION is defined? I think we should
>> document this.
>>
>> https://codereview.chromium.org/246633012/
>>
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[jfroy, 2014-05-05 21:33:39]

cpu@, git cl owners says you own this config file. Can you take a look?

[cpu_(ooo_6.6-7.5), 2014-05-08 17:27:35]

not exactly, see third_party\owners file.

[jfroy, 2014-05-08 17:42:34]

On 2014/05/08 17:27:35, cpu wrote:
> not exactly, see third_party\owners file.

Oops, sorry!

[jfroy, 2014-05-09 20:41:04]

The CQ bit was checked by jfroy@chromium.org

[commit-bot: I haz the power, 2014-05-09 20:42:12]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/jfroy@chromium.org/246633012/100001

[jfroy, 2014-05-09 20:57:26]

The CQ bit was checked by jfroy@chromium.org

[commit-bot: I haz the power, 2014-05-09 20:59:08]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/jfroy@chromium.org/246633012/120001

[commit-bot: I haz the power, 2014-05-11 00:15:20]

Change committed as 269646