OBSOLETE: Move (most of) chrome/browser/chromeos/drive into components/drive.

OBSOLETE: This CL has been replaced by smaller CLs that have recently concluded with crrev.com/1314803004.
---------

Move (most of) chrome/browser/chromeos/drive into components/drive.

Note that some files under components/drive have not been simply
moved, but had to undergo small changes (in addition to the mostly
mechanical change of include paths):

- Hiding small chunks of code behind if defined(OS_CHROMEOS):
  file_cache.cc
  file_system/download_operation_unittest.cc
  test_util.h

- Hiding small chunks of code behind if defines(OS_POSIX):
  file_cache.cc

- Removing unneeded includes (to limit dependencies in
  components/drive/DEPS):
  components/drive/file_cache.cc
  components/drive/file_change.h
  components/drive/file_system_core_util.cc

As with the previous change (crrev.com/1190203002) tests for now continue to
depend on the browser and be executed as part of the browser's unit tests.

Things left behind in chrome/browser/chromeos/drive are not needed by
drive::FileSystem and therefore there is no pressing need to componentize them.

Test steps:
1. Verify that things still build via GYP (and unit tests pass).
    $ GYP_DEFINES="use_goma=1 gomadir=... chromeos=1" gclient sync
    $ ninja -C out/Debug -j 150 chrome unit_tests \
                 interactive_ui_tests browser_tests drive
    $ out/Debug/unit_tests

2. Verify that things still build via GN.
    $ gn gen out/Default --args='target_os="chromeos" use_goma=true'
    $ ninja -C out/Default -j 150 chrome unit_tests \
                 interactive_ui_tests browser_tests components/drive

TEST=Please see "Test steps" above.
BUG=257943, 498951

[Łukasz Anforowicz, 2015-07-01 23:07:39]

lukasza@chromium.org changed reviewers:
+ hashimoto@chromium.org, mtomasz@chromium.org

[Łukasz Anforowicz, 2015-07-01 23:07:41]

Ryo and Tomasz - could you please take a look?

While crrev.com/1190203002 (moving chrome/browser/drive into components/drive)
is waiting for one final LGTM, I thought that it might be useful to kick-off
reviewing of the current CL - crrev.com/1215503010 (moving
chrome/browser/chromeos/drive into components/drive).  

The current CL builds and passes tests from the description on my local box, but
"git cl try" didn't help with building and testing on other platforms (because
of the dependency on crrev.com/1190203002).  Even with not-quite-complete
testing, I hope that kicking off the reviews would be the most efficient use of
everyone's time (especially since my part of the world will stop all work during
the long 4th-of-July weekend :-).

Thanks,

Lukasz

[mtomasz, 2015-07-02 07:36:34]

Hard to take a grasp on such a huge CL. I took a quick look and lgtm with some
nits. I'm not a DEPS/gyp/gn master, so please make sure someone else checks
that.

https://codereview.chromium.org/1215503010/diff/1/components/drive/file_syste...
File components/drive/file_system/download_operation_unittest.cc (right):

https://codereview.chromium.org/1215503010/diff/1/components/drive/file_syste...
components/drive/file_system/download_operation_unittest.cc:31: const int64
kMinFreeSpaceInBytes = 512 * 1LL << 20;
It's quite cryptic. Can we use 512 * 1024 * 1024 instead? Also, is this value
correct for other OS'es?

https://codereview.chromium.org/1215503010/diff/1/components/drive/test_util.h
File components/drive/test_util.h (right):

https://codereview.chromium.org/1215503010/diff/1/components/drive/test_util....
components/drive/test_util.h:34: const int64 kMinFreeSpaceInBytes = 10 * 512 *
1LL << 20;
nit: ditto?

[hashimoto, 2015-07-06 06:35:46]

Could you split this CL in to a number of chunks (e.g. moving
c/b/chromeos/drive/sync)?
It seems this is a bit too large to be a single CL.

[Łukasz Anforowicz, 2015-07-20 22:57:08]

On 2015/07/06 06:35:46, hashimoto wrote:
> Could you split this CL in to a number of chunks (e.g. moving
> c/b/chromeos/drive/sync)?

This is tricky - we don't want to have a components/drive -> chrome/...
dependencies, so we can only move a "self-contained" subset (i.e. "leaves" of
the dependency tree).  This means that moving only c/b/chromeos/drive/sync is
not possible.  OTOH, I can try moving drive.proto first and then "leaves" (i.e.
drive_pref_names, file_change, file_errors, local_file_reader) and then try
evaluating the experience + try forecasting how many "small" CLs are remaining
at this point (I say "small" in quotes, because even just moving a single
"drive.proto" file ends up touching 50+ files).  

> It seems this is a bit too large to be a single CL.

Let me try moving just drive.proto to see if smaller changes can be done more
quickly (crrev.com/1243853002).

[hashimoto, 2015-07-21 11:16:20]

On 2015/07/20 22:57:08, Łukasz Anforowicz wrote:
> On 2015/07/06 06:35:46, hashimoto wrote:
> > Could you split this CL in to a number of chunks (e.g. moving
> > c/b/chromeos/drive/sync)?
> 
> This is tricky - we don't want to have a components/drive -> chrome/...
> dependencies, so we can only move a "self-contained" subset (i.e. "leaves" of
> the dependency tree).  This means that moving only c/b/chromeos/drive/sync is
> not possible.  OTOH, I can try moving drive.proto first and then "leaves"
(i.e.
> drive_pref_names, file_change, file_errors, local_file_reader) and then try
> evaluating the experience + try forecasting how many "small" CLs are remaining
> at this point (I say "small" in quotes, because even just moving a single
> "drive.proto" file ends up touching 50+ files).  
> 
> > It seems this is a bit too large to be a single CL.
> 
> Let me try moving just drive.proto to see if smaller changes can be done more
> quickly (crrev.com/1243853002).

Thank you for making a smaller CL.
It's much more doable to review a CL which changes 50+ files (40+ .cc files with
1-line include change, and a few GYP/GN changes) than this CL (here even
Rietveld says "Patch set is too large to download")!
It should make the review process less error-prone.