Add in boost_tuple into third-party for use by googlemock in windows.

Add in boost_tuple into third-party for use by googlemock in windows.

[awong, 2009-05-15 10:35:01]

This CL is split out from http://codereview.chromium.org/113163 which no longer
will load after I uploaded the boost_tuple implemenation files.

Currently, the CL only has the README.chromium and the boost.gyp file.  The
other change that this CL should include is the unzipped boost_tuple files
(http://googlemock.googlecode.com/files/boost_tr1_tuple_1_36_0.zip).

I'm thinking that since uploading those files to rietveld crashed it last time,
maybe this time, we just review the boost.gyp and README.chromium, and then I
unzip these files, and push to SVN under this CL without reuploading the patch.

Sound okay?

The three possible sticky points here are:
  a) did I graft the files onto the tree correctly? (see README.chromium for
details)
  b) including this does suddenly make tr1::tuple available in our codebase. Is
there a way to make depending on this outside of a unittest invalid? 
  c) Should we just pull all of boost via a normal SVN dependency and avoid
having to graft the files into the tree?  Upside: easier maintenance. Downside:
all of boost shows up.

-Albert

[sgk, 2009-05-15 13:37:29]

http://codereview.chromium.org/115398/diff/1/2
File third_party/boost/README.chromium (right):

http://codereview.chromium.org/115398/diff/1/2#newcode12
Line 12: The gyp project file is hand written to make this file available for
windows
Drop this paragraph in favor of appropriate comment(s) in the boost.gyp file
itself.  This is the sort of paragraph in a separate file that someone will
probably forget to update when we do, some day, enable it for Mac and/or Linux.

http://codereview.chromium.org/115398/diff/1/3
File third_party/boost/boost.gyp (right):

http://codereview.chromium.org/115398/diff/1/3#newcode12
Line 12: 'sources': [
I'd list all of the files here, and then exclude all of them on Mac and Linux. 
That'll make life much simpler if/when we start enabling parts of this on those
platforms.  You should be able to do this with an exclude pattern like so
(modulo off-the-top-of-my-head mistakes):

['OS!="win"', {
  # We don't use this yet on Mac and Linux,
  # so exclude everything from the source list.
  'sources/': [
    ['exclude': '.*'],
  ],
}]

[M-A Ruel, 2009-05-15 14:32:16]

What? Boost? The answer is simple: no.

I'm *not* in favor to add 6 megs of include headers in the tree for only tuple
support, especially that we already have a tuple class.

[Nicolas Sylvain, 2009-05-15 14:40:29]

LGTM with my comment and sgk's comments.

As for maruel's comment : It made me sad too. gmock seems to be a lot of
troubles (svn:external, then ugly dependencies).  Have we at least considered
using something else? Or not using it at all?

http://codereview.chromium.org/115398/diff/1/2
File third_party/boost/README.chromium (right):

http://codereview.chromium.org/115398/diff/1/2#newcode3
Line 3: http://googlemock.googlecode.com/files/boost_tr1_tuple_1_36_0.zip
Can you add a line that says what the license is.

[awong, 2009-05-16 03:14:46]

Here is an updated version of the files.

I moved boost down one directory, cause it felt cleaner.  I also took John's
suggestion for how to isolate the critical subset of files, and kept only those.
 Still a lot of files, but much better.  I tested it locally on my desktop and
the subset does work for my unittest, which uses the gmock-action api (the one
that needs tr1).

maurel: I don't know how conditionally disable the boost inclusion based on the
msvc version.  If you have a suggestion, I'm all ears.  Currently, I expect that
any system install of tr1 will shadow this version because of the <> in the
include which, IIRC, usually means search system headers first on most
compilers.

REMINDER: this CL should be considered to include the 390 boost files.  I'm just
not uploading them to rietveld since the size causes the review page to stop
loading.

http://codereview.chromium.org/115398/diff/1/2
File third_party/boost/README.chromium (right):

http://codereview.chromium.org/115398/diff/1/2#newcode3
Line 3: http://googlemock.googlecode.com/files/boost_tr1_tuple_1_36_0.zip
On 2009/05/15 14:40:29, Nicolas Sylvain wrote:
> Can you add a line that says what the license is.

Done.

http://codereview.chromium.org/115398/diff/1/2#newcode12
Line 12: The gyp project file is hand written to make this file available for
windows
On 2009/05/15 13:37:29, sgk wrote:
> Drop this paragraph in favor of appropriate comment(s) in the boost.gyp file
> itself.  This is the sort of paragraph in a separate file that someone will
> probably forget to update when we do, some day, enable it for Mac and/or
Linux.

Done.  We will probably never enable this in mac/linux since they already have
tr1.

http://codereview.chromium.org/115398/diff/1/3
File third_party/boost/boost.gyp (right):

http://codereview.chromium.org/115398/diff/1/3#newcode12
Line 12: 'sources': [
On 2009/05/15 13:37:29, sgk wrote:
> I'd list all of the files here, and then exclude all of them on Mac and Linux.

> That'll make life much simpler if/when we start enabling parts of this on
those
> platforms.  You should be able to do this with an exclude pattern like so
> (modulo off-the-top-of-my-head mistakes):
> 
> ['OS!="win"', {
>   # We don't use this yet on Mac and Linux,
>   # so exclude everything from the source list.
>   'sources/': [
>     ['exclude': '.*'],
>   ],
> }]

Swapped. running through try server.  Had to leave the direct_dependent_settings
field though.

[awong, 2009-05-18 22:29:21]

Got the gyp file working.

Deferring dealing with the include paths until the gmock checkin.

[sgk, 2009-05-19 01:14:28]

Stripping down what we're going to check in seems good.  As it's written now,
though, this is a do-nothing target on Mac and Linux.  It'll be more efficient
(and arguably simpler) to have the other code that uses this only express a
dependency on this target if it's windows.

http://codereview.chromium.org/115398/diff/1011/1013
File third_party/boost/boost.gyp (right):

http://codereview.chromium.org/115398/diff/1011/1013#newcode420
Line 420: 'sources/': [['exclude', '.*']],
If this whole target is really for windows only, then it probably makes the most
sense to just eliminate this entire condition.  Make the whole target windows
only, and put the dependencies on it in 'conditions' blocks in the other .gyp
files that reference it.  Otherwise, we're just complicating and slowing down
the mac and linux builds to no purpose.

[awong, 2009-05-19 01:34:44]

How's this look?

[M-A Ruel, 2009-05-19 13:56:45]

lgtm for me, I didn't look at the gyp file.

http://codereview.chromium.org/115398/diff/1017/9
File third_party/boost/README.chromium (right):

http://codereview.chromium.org/115398/diff/1017/9#newcode26
Line 26: In comparison the full exapnsion of boost_tr1_tuple_1_36_0.zip is 1097
files
expansion

[awong, 2009-05-19 18:38:59]

http://codereview.chromium.org/115398/diff/1017/9
File third_party/boost/README.chromium (right):

http://codereview.chromium.org/115398/diff/1017/9#newcode26
Line 26: In comparison the full exapnsion of boost_tr1_tuple_1_36_0.zip is 1097
files
On 2009/05/19 13:56:46, M-A wrote:
> expansion

Fixed.

[zhanyong.wan, 2009-05-19 19:03:01]

Thanks for the hard work, Albert.  I hope it will turn out worthwhile for the
Chromium team. :-)

http://codereview.chromium.org/115398/diff/15/1019
File third_party/boost/README.chromium (right):

http://codereview.chromium.org/115398/diff/15/1019#newcode16
Line 16: #include <tr1/tuple.h>
According to Boost's documentation, to use its tuple implementation we should:

- add boost_root/boost/tr1/tr1;boost_root at the beginning of the include search
path.
- #include <tuple>

Could you change this to #include <tuple>?

http://codereview.chromium.org/115398/diff/15/1019#newcode17
Line 17: 4) Run this file through the Visual C++ preprocessor (use the /P flag
to CL).
Could you provide the command line flags you used to run cl?  In particular, the
include search path is critical.

Ideally, the flags should match what you use in the real build, in case Boost
tr1 tuple does conditional #includes.

[awong, 2009-05-19 20:07:56]

Thanks for the comments!

Updated the readme again.

http://codereview.chromium.org/115398/diff/15/1019
File third_party/boost/README.chromium (right):

http://codereview.chromium.org/115398/diff/15/1019#newcode16
Line 16: #include <tr1/tuple.h>
On 2009/05/19 19:03:02, zhanyong.wan wrote:
> According to Boost's documentation, to use its tuple implementation we should:
> 
> - add boost_root/boost/tr1/tr1;boost_root at the beginning of the include
search
> path.
> - #include <tuple>
> 
> Could you change this to #include <tuple>?

Good to know.  Changed the instructions to match.

http://codereview.chromium.org/115398/diff/15/1019#newcode17
Line 17: 4) Run this file through the Visual C++ preprocessor (use the /P flag
to CL).
On 2009/05/19 19:03:02, zhanyong.wan wrote:
> Could you provide the command line flags you used to run cl?  In particular,
the
> include search path is critical.
> 
> Ideally, the flags should match what you use in the real build, in case Boost
> tr1 tuple does conditional #includes.

I didn't actually run CL manually, but did it via the visual studio so as to
pickup the relevant project settings and includes.  I've updated the
documentation to match that.

[zhanyong.wan, 2009-05-19 20:16:08]

Only one minor problem now.  Thanks, Albert.

http://codereview.chromium.org/115398/diff/18/19
File third_party/boost/README.chromium (right):

http://codereview.chromium.org/115398/diff/18/19#newcode16
Line 16: #include <tuple.h>
It should be <tuple>, not <tuple.h>.

[awong, 2009-05-19 20:24:23]

http://codereview.chromium.org/115398/diff/18/19
File third_party/boost/README.chromium (right):

http://codereview.chromium.org/115398/diff/18/19#newcode16
Line 16: #include <tuple.h>
On 2009/05/19 20:16:08, zhanyong.wan wrote:
> It should be <tuple>, not <tuple.h>.

Done.

[zhanyong.wan, 2009-05-19 20:27:46]

Looks good!

[awong, 2009-05-23 01:39:11]

Okay, I submitted this as r16829.  If anyone wants changes, let me know and I'll
open another CL.

Now, onto the real goal: gmock.