Transfer the C++03 move-only type emulation into base/move.h and also make ScopedVector move-only.

Transfer the C++03 move-only type emulation into base/move.h and also make ScopedVector move-only.

Also:
  * Add a lot of documentation explaining what this macro does.
  * Change the implementation of RValue so it cannot be instantiated.

The change to always use RValue& makes for more efficent code in debug builds.  Looking at the disassembly for a simple use case (calling a function with one parameter), it removes the creation of one temporary.

BUG=96118
TEST=new unittests. exist code still compiles.

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=118388

[kim.grasman, 2012-01-18 08:27:19]

Comment nit.

http://codereview.chromium.org/9207021/diff/16001/base/move.h
File base/move.h (right):

http://codereview.chromium.org/9207021/diff/16001/base/move.h#newcode98
base/move.h:98: // IMPLEMENTATION SUBTLTIES WITH RValue
Nice spelling, but I think "SUBTLETIES" is more established.

[awong, 2012-01-18 23:21:04]

Also ran through spell check. :)

http://codereview.chromium.org/9207021/diff/16001/base/move.h
File base/move.h (right):

http://codereview.chromium.org/9207021/diff/16001/base/move.h#newcode98
base/move.h:98: // IMPLEMENTATION SUBTLTIES WITH RValue
On 2012/01/18 08:27:20, kim.grasman wrote:
> Nice spelling, but I think "SUBTLETIES" is more established.

Thanks. Fixed!

[darin (slow to review), 2012-01-19 05:17:30]

Sorry for the slow reply.

http://codereview.chromium.org/9207021/diff/13009/base/memory/scoped_vector.h
File base/memory/scoped_vector.h (right):

http://codereview.chromium.org/9207021/diff/13009/base/memory/scoped_vector.h...
base/memory/scoped_vector.h:19: MOVE_ONLY_TYPE_FOR_CPP_03(ScopedVector);
I think it might improve readability of the code to pass RValue as a parameter
to this macro.  That way, if you search within this file for RValue you will hit
that.  Otherwise, seeing RValue down below might leave one quite perplexed as to
where it gets defined.

http://codereview.chromium.org/9207021/diff/13009/base/move.h
File base/move.h (right):

http://codereview.chromium.org/9207021/diff/13009/base/move.h#newcode13
base/move.h:13: // a "move-only" type.  Unlike, this macro should be the first
line in a
Unlike [what]?

http://codereview.chromium.org/9207021/diff/13009/base/move.h#newcode139
base/move.h:139: // impossible to make an function "void Foo(scoped_ptr<Parent>
p)" accept a
this makes me a little sad, but ok :)

[awong, 2012-01-19 20:13:12]

Fixed and submitting.

Thanks for the rush review!

http://codereview.chromium.org/9207021/diff/13009/base/move.h
File base/move.h (right):

http://codereview.chromium.org/9207021/diff/13009/base/move.h#newcode13
base/move.h:13: // a "move-only" type.  Unlike, this macro should be the first
line in a
On 2012/01/19 05:17:30, darin wrote:
> Unlike [what]?

Done.   (Unlike DISALLOW_COPY_AND_ASSIGN)

http://codereview.chromium.org/9207021/diff/13009/base/move.h#newcode139
base/move.h:139: // impossible to make an function "void Foo(scoped_ptr<Parent>
p)" accept a
On 2012/01/19 05:17:30, darin wrote:
> this makes me a little sad, but ok :)

Yes.  Makes me sad too.

This stack overflow has a good walk through of why.  sergeyu@ was thinking of
adding a PassAs<>() function to smooth this over a bit.  You may see a review
from him at some point for it.

http://stackoverflow.com/questions/2365031/does-hinnants-unique-ptr-implement...

[commit-bot: I haz the power, 2012-01-19 20:14:49]

No LGTM from any valid reviewer yet. Even if a LGTM may have been
provided, it was from a non-committer or a provisional committer
and not a full committer.
See http://www.chromium.org/getting-involved/become-a-committer
Note that this has nothing to do with OWNERS files.

[awong, 2012-01-19 20:17:52]

Bah. I got an "LGTM w/ those nits picked" from darin over e-mail, but apparently
it didn't show up on the review site.

I'm going to just commit -f.

On 2012/01/19 20:13:12, awong wrote:
> Fixed and submitting.
> 
> Thanks for the rush review!
> 
> http://codereview.chromium.org/9207021/diff/13009/base/move.h
> File base/move.h (right):
> 
> http://codereview.chromium.org/9207021/diff/13009/base/move.h#newcode13
> base/move.h:13: // a "move-only" type.  Unlike, this macro should be the first
> line in a
> On 2012/01/19 05:17:30, darin wrote:
> > Unlike [what]?
> 
> Done.   (Unlike DISALLOW_COPY_AND_ASSIGN)
> 
> http://codereview.chromium.org/9207021/diff/13009/base/move.h#newcode139
> base/move.h:139: // impossible to make an function "void
Foo(scoped_ptr<Parent>
> p)" accept a
> On 2012/01/19 05:17:30, darin wrote:
> > this makes me a little sad, but ok :)
> 
> Yes.  Makes me sad too.
> 
> This stack overflow has a good walk through of why.  sergeyu@ was thinking of
> adding a PassAs<>() function to smooth this over a bit.  You may see a review
> from him at some point for it.
> 
>
http://stackoverflow.com/questions/2365031/does-hinnants-unique-ptr-implement...

[darin (slow to review), 2012-01-19 21:23:41]

LGTM for realz

[commit-bot: I haz the power, 2012-01-19 21:24:42]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/ajwong@chromium.org/9207021/14003

[commit-bot: I haz the power, 2012-01-19 23:22:41]

Try job failure for 9207021-14003 (retry) on mac_rel for step "browser_tests".
It's a second try, previously, step "browser_tests" failed.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=mac_rel&nu...