Move chromium-specific files from leveldb to chromium.

Move chromium-specific files from leveldb's repository to chromium, stage 2.

Stages:
1. Remove chromium files from leveldb repo. [Done]
2. Put chromium leveldb files in third_party/leveldatabase and leveldb in third_party/leveldatabase/src.  Update leveldb references in chrome to use third_party/leveldatabase/src instead of third_party/leveldb.  Leave third_party/leveldb so it can be referred to by WebCore.gyp.
3. Change third_party/WebKit/Source/WebCore/WebCore.gyp/WebCore.gyp to use third_party/leveldatabase.
4. Delete third_party/leveldb
Eventually: 5. Rename third_party/leveldatabase back to third_party/leveldb

We can't avoid the leveldb-checkout duplication by just changing third_party/leveldb from an unversioned to a versioned directory. Doing so would break everyone's gclient sync update.

BUG=89378
TEST=indexeddb layout and browser tests


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=96915

[dgrogan, 2011-07-27 23:57:02]

port_chromium.* and env_chromium were copied without modification.

[hans, 2011-07-28 15:25:45]

On 2011/07/27 23:57:02, dgrogan wrote:
> port_chromium.* and env_chromium were copied without modification.

LGTM, tanks for fixing! Try jobs?

[hans, 2011-07-28 15:36:16]

On 2011/07/28 15:25:45, hans wrote:
> On 2011/07/27 23:57:02, dgrogan wrote:
> > port_chromium.* and env_chromium were copied without modification.
> 
> LGTM, tanks for fixing! Try jobs?

(and by tanks, I mean thanks obviously :)

[dgrogan, 2011-08-06 00:12:14]

Hans, could you take a look at this?

[dgrogan, 2011-08-06 01:47:09]

On 2011/08/06 00:12:14, dgrogan wrote:
> Hans, could you take a look at this?

With the trybots at 0/3 (though linux is just flakiness) this isn't ready for a
proper review.

But I am open to suggestions on how to do this better.  It's obviously awkward
to have 2 simultaneous checkouts of leveldb.

[hans, 2011-08-08 15:06:40]

On 2011/08/06 01:47:09, dgrogan wrote:
> On 2011/08/06 00:12:14, dgrogan wrote:
> > Hans, could you take a look at this?
> 
> With the trybots at 0/3 (though linux is just flakiness) this isn't ready for
a
> proper review.

What are the errors on the non-Linux bots?

> 
> But I am open to suggestions on how to do this better.  It's obviously awkward
> to have 2 simultaneous checkouts of leveldb.

I don't see a better way... it's the WebKit that's the problem, right? That we
can't simultaneously change the location of leveldb Chromium-side, and update
the location in the gyp-file in WebKit..

Who would be a good person to ask about finding a simpler way of doing this?
Evan Martin, maybe?

[hans, 2011-08-08 15:06:53]

http://codereview.chromium.org/7522008/diff/4003/third_party/leveldatabase/le...
File third_party/leveldatabase/leveldb.gyp (right):

http://codereview.chromium.org/7522008/diff/4003/third_party/leveldatabase/le...
third_party/leveldatabase/leveldb.gyp:164: 'util/testutil.h',
Shouldn't these be 'src/util/...'?

[dgrogan, 2011-08-09 06:27:35]

On 2011/08/08 15:06:40, hans wrote:
> On 2011/08/06 01:47:09, dgrogan wrote:
> > On 2011/08/06 00:12:14, dgrogan wrote:
> > > Hans, could you take a look at this?
> > 
> > With the trybots at 0/3 (though linux is just flakiness) this isn't ready
for
> a
> > proper review.
> 
> What are the errors on the non-Linux bots?

The windows stuff is standard windows compile annoyances.  I'm not too worried
about it.

The mac stuff is more concerning.  The error is:
________ running '/usr/bin/python src/build/gyp_chromium' in
'/b/build/slave/mac/build'
Updating projects from gyp files...
Traceback (most recent call last):
[snip]
  File "/b/build/slave/mac/build/src/tools/gyp/pylib/gyp/xcodeproj_file.py",
line 460, in EnsureNoIDCollisions
    str(other._properties), self._properties['rootObject'].Name())
KeyError: 'Duplicate ID F08934AC668C13F7EE983DA6, objects "{\'children\':
[<PBXReferenceProxy \'libleveldatabase.a\' at 0x10294eb10>], \'name\':
\'Products\', \'sourceTree\': \'<group>\'}" and "{\'children\':
[<PBXReferenceProxy \'libleveldb.a\' at 0x10294b410>, <PBXReferenceProxy
\'libleveldb_testutil.a\' at 0x10294b490>, <PBXReferenceProxy
\'leveldb_arena_test\' at 0x10294b590>, <PBXReferenceProxy
\'leveldb_cache_test\' at 0x10294b810>, <PBXReferenceProxy
\'leveldb_coding_test\' at 0x10294ba10>, <PBXReferenceProxy
\'leveldb_corruption_test\' at 0x10294b910>, <PBXReferenceProxy
\'leveldb_crc32c_test\' at 0x10294b710>, <PBXReferenceProxy \'leveldb_db_bench\'
at 0x10294b610>, <PBXReferenceProxy \'leveldb_db_test\' at 0x10294bc10>,
<PBXReferenceProxy \'leveldb_dbformat_test\' at 0x10294ba90>, <PBXReferenceProxy
\'leveldb_env_test\' at 0x10294b890>, <PBXReferenceProxy
\'leveldb_filename_test\' at 0x10294b790>, <PBXReferenceProxy
\'leveldb_log_test\' at 0x10294bb90>, <PBXReferenceProxy
\'leveldb_skiplist_test\' at 0x10294b690>, <PBXReferenceProxy
\'leveldb_table_test\' at 0x10294b990>, <PBXReferenceProxy
\'leveldb_version_edit_test\' at 0x10294b510>, <PBXReferenceProxy
\'leveldb_write_batch_test\' at 0x10294bb10>], \'name\': \'Products\',
\'sourceTree\': \'<group>\'}" in "content"'
Error: /usr/bin/python src/build/gyp_chromium in /b/build/slave/mac/build
returned 1
elapsedTime=35.606840
program finished with exit code 2

There's some problem concerning the similarity between leveldb.a and
leveldatabase.a.

> 
> > 
> > But I am open to suggestions on how to do this better.  It's obviously
awkward
> > to have 2 simultaneous checkouts of leveldb.
> 
> I don't see a better way... it's the WebKit that's the problem, right? That we
> can't simultaneously change the location of leveldb Chromium-side, and update
> the location in the gyp-file in WebKit..
> 
> Who would be a good person to ask about finding a simpler way of doing this?
> Evan Martin, maybe?

[dgrogan, 2011-08-09 06:27:56]

http://codereview.chromium.org/7522008/diff/4003/third_party/leveldatabase/le...
File third_party/leveldatabase/leveldb.gyp (right):

http://codereview.chromium.org/7522008/diff/4003/third_party/leveldatabase/le...
third_party/leveldatabase/leveldb.gyp:164: 'util/testutil.h',
On 2011/08/08 15:06:53, hans wrote:
> Shouldn't these be 'src/util/...'?

Yes.  Instead of fixing it I deleted this target for now.

[dgrogan, 2011-08-09 06:31:45]

Brett,

Could you look over the plan outlined in the issue Description?  This is the
best we've come up with, but it's awkward to have 2 simultaneous leveldb
checkouts.  Do you have advice or ideas how this migration could be done
simpler?

Thanks a bunch.

[brettw, 2011-08-09 18:26:06]

Sorry, I have no brilliant idea to solve the problem. Your plan sounds fine.

[dgrogan, 2011-08-10 03:38:32]

Mark,

Could you advise on how to work around the EnsureNoIDCollisions error raised in
gyp/xcodeproj_file.py copied below?  I'm trying to temporarily include two
copies of a third_party library in the chromium build.  What can I do to make
them just different enough to not trigger an ID collision?

Thanks for your help.

________ running '/usr/bin/python src/build/gyp_chromium' in
'/b/build/slave/mac/build'
Updating projects from gyp files...
Traceback (most recent call last):
  File "src/build/gyp_chromium", line 170, in <module>
    sys.exit(gyp.main(args))
  File "/b/build/slave/mac/build/src/tools/gyp/pylib/gyp/__init__.py", line 472,
in main
    generator.GenerateOutput(flat_list, targets, data, params)
  File "/b/build/slave/mac/build/src/tools/gyp/pylib/gyp/generator/xcode.py",
line 1201, in GenerateOutput
    xcode_target_to_target_dict)
  File "/b/build/slave/mac/build/src/tools/gyp/pylib/gyp/generator/xcode.py",
line 416, in Finalize2
    self.project_file.EnsureNoIDCollisions()
  File "/b/build/slave/mac/build/src/tools/gyp/pylib/gyp/xcodeproj_file.py",
line 460, in EnsureNoIDCollisions
    str(other._properties), self._properties['rootObject'].Name())
KeyError: 'Duplicate ID F08934AC668C13F7EE983DA6, objects "{\'children\':
[<PBXReferenceProxy \'libleveldatabase.a\' at 0x102995310>], \'name\':
\'Products\', \'sourceTree\': \'<group>\'}" and "{\'children\':
[<PBXReferenceProxy \'libleveldb.a\' at 0x102990bd0>, <PBXReferenceProxy
\'libleveldb_testutil.a\' at 0x102990c50>, <PBXReferenceProxy
\'leveldb_arena_test\' at 0x102990d50>, <PBXReferenceProxy
\'leveldb_cache_test\' at 0x102990fd0>, <PBXReferenceProxy
\'leveldb_coding_test\' at 0x102991210>, <PBXReferenceProxy
\'leveldb_corruption_test\' at 0x102991110>, <PBXReferenceProxy
\'leveldb_crc32c_test\' at 0x102990ed0>, <PBXReferenceProxy \'leveldb_db_bench\'
at 0x102990dd0>, <PBXReferenceProxy \'leveldb_db_test\' at 0x102991410>,
<PBXReferenceProxy \'leveldb_dbformat_test\' at 0x102991290>, <PBXReferenceProxy
\'leveldb_env_test\' at 0x102991090>, <PBXReferenceProxy
\'leveldb_filename_test\' at 0x102990f50>, <PBXReferenceProxy
\'leveldb_log_test\' at 0x102991390>, <PBXReferenceProxy
\'leveldb_skiplist_test\' at 0x102990e50>, <PBXReferenceProxy
\'leveldb_table_test\' at 0x102991190>, <PBXReferenceProxy
\'leveldb_version_edit_test\' at 0x102990cd0>, <PBXReferenceProxy
\'leveldb_write_batch_test\' at 0x102991310>], \'name\': \'Products\',
\'sourceTree\': \'<group>\'}" in "content"'
Error: /usr/bin/python src/build/gyp_chromium in /b/build/slave/mac/build
returned 1
elapsedTime=33.250413
program finished with exit code 2

[Mark Mentovai, 2011-08-10 12:55:43]

dgrogan@chromium.org wrote:
> Mark,
>
> Could you advise on how to work around the EnsureNoIDCollisions error raised
in
> gyp/xcodeproj_file.py copied below?  I'm trying to temporarily include two
> copies of a third_party library in the chromium build.  What can I do to
> make them just different enough to not trigger an ID collision?

You don’t do that.

How identical are the two copies? At a minimum, the target names and
product names should be different. Otherwise, it isn’t only an ID
collision, but it’s totally ambiguous as to which target is supposed
to produce the output.

[dgrogan, 2011-08-10 20:38:28]

On 2011/08/10 12:55:43, Mark Mentovai wrote:
> mailto:dgrogan@chromium.org wrote:
> > Mark,
> >
> > Could you advise on how to work around the EnsureNoIDCollisions error raised
> in
> > gyp/xcodeproj_file.py copied below?  I'm trying to temporarily include
two
> > copies of a third_party library in the chromium build.  What can I do
to
> > make them just different enough to not trigger an ID collision?
> 
> You don’t do that.

Which "that"?

> How identical are the two copies?

They are 11 revisions apart, small but nontrivial source changes.  The object
code should be different.

> At a minimum, the target names and
> product names should be different. Otherwise, it isn’t only an ID
> collision, but it’s totally ambiguous as to which target is supposed
> to produce the output.

The target names are different ('leveldatabase' vs 'leveldb'), and no
product_name is specified.  Do I need to change something else?  How are the IDs
generated?  What does xcode use them for?

[dgrogan, 2011-08-13 00:07:49]

Hans, could you take another look at this?  It seems to be working now.

I changed env_chromium.cc to use chromium's logging instead of leveldb's. 
Mostly because leveldb's PosixLogger class uses gettimeofday, which isn't
available on windows.  LevelDB messages are logged with VLOG(5), 5 arbitrarily
picked because it seems relatively high.

Compare env_chromium.cc with that in patch set 3 to see the differences.

[hans, 2011-08-15 11:27:31]

On 2011/08/13 00:07:49, dgrogan wrote:
> Hans, could you take another look at this?  It seems to be working now.
> 
> I changed env_chromium.cc to use chromium's logging instead of leveldb's. 
> Mostly because leveldb's PosixLogger class uses gettimeofday, which isn't
> available on windows.  LevelDB messages are logged with VLOG(5), 5 arbitrarily
> picked because it seems relatively high.
> 
> Compare env_chromium.cc with that in patch set 3 to see the differences.

LGTM.

What changed to make the Mac bot happy?

[dgrogan, 2011-08-15 18:07:44]

On 2011/08/15 11:27:31, hans wrote:
> On 2011/08/13 00:07:49, dgrogan wrote:
> > Hans, could you take another look at this?  It seems to be working now.
> > 
> > I changed env_chromium.cc to use chromium's logging instead of leveldb's. 
> > Mostly because leveldb's PosixLogger class uses gettimeofday, which isn't
> > available on windows.  LevelDB messages are logged with VLOG(5), 5
arbitrarily
> > picked because it seems relatively high.
> > 
> > Compare env_chromium.cc with that in patch set 3 to see the differences.
> 
> LGTM.
> 
> What changed to make the Mac bot happy?

Looks like changing the new gyp file name from leveldb.gyp to leveldatabase.gyp.
 But that seems weird so I'm not sure, and haven't heard back from Mentovai.

[Mark Mentovai, 2011-08-15 18:22:16]

Why’s everyone thinkin’ I’m all AWOL? I see my reply to this thread on
the codereview.

Yes, this can be caused by having two identically-named .gyp files in
different directories.

dgrogan@chromium.org wrote:
> On 2011/08/15 11:27:31, hans wrote:
>>
>> On 2011/08/13 00:07:49, dgrogan wrote:
>> > Hans, could you take another look at this?  It seems to be working now.
>> >
>> > I changed env_chromium.cc to use chromium's logging instead of
>> > leveldb's.
>> > Mostly because leveldb's PosixLogger class uses gettimeofday, which
>> > isn't
>> > available on windows.  LevelDB messages are logged with VLOG(5), 5
>
> arbitrarily
>>
>> > picked because it seems relatively high.
>> >
>> > Compare env_chromium.cc with that in patch set 3 to see the differences.
>
>> LGTM.
>
>> What changed to make the Mac bot happy?
>
> Looks like changing the new gyp file name from leveldb.gyp to
> leveldatabase.gyp.
>  But that seems weird so I'm not sure, and haven't heard back from Mentovai.
>
> http://codereview.chromium.org/7522008/
>

[commit-bot: I haz the power, 2011-08-16 11:21:00]

Change committed as 96915