Import SQLite 3.7.6.3.

Import SQLite 3.7.6.3.

Ran through the import script in third_party/sqlite/README.Chromium,
including the SQLite test suite.  A few minor errors under
oserror-1.1.[123], I suspect my Linux reference machine is becoming
outdated.

SQLite changes are going to be hard/impossible to review.  It's a year
and a half of changes, might as well be a new package.

Removed patches which are no longer necessary:
  icu-regepx.patch is already present
  safe-tolower.patch is already present
  attach-integer.patch no longer relevant

For attach-integer.patch, the code has been refactored.  The failure
in http://crbug.com/38745 no longer occurs (ATTACH 12345 as db;
attaches the SQLite database "12345" rather than crashing).

Tweak test.patch for modern era.

Adjusted webdb.patch for new flag.

Clean up fts3.patch.  fts3 has been substantially refactored, so much
of the patch no longer applies.  Reviewed fts3_write.c and many of the
patches are now handled appropriately.  I found one issue in fts3.c,
and I'm still reviewing it.  This will be ongoing.

Disabled WAL (Write Ahead Log) and fts4 support for now as unknown
unknowns.

BUG=71731
TEST=*EVERYTHING* continues to work.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=86616

[Scott Hess - ex-Googler, 2011-05-23 22:22:41]

Apologies.

I ran things through the trybots earlier, on CL at
http://codereview.chromium.org/7042020/ .  When I went to re-upload, it ...
didn't work, or something, lots of odd stuff was going on with completely
unrelated files.  So I spun off a new CL.  So it's possible I won't be able to
easily upload changes (though I'll keep my git repo static for now).

The 3.7.6.2 CL had only a small WAL fix added for 3.7.6.3, so it shouldn't
impact the eventual results of the current try jobs.

For future reference, http://codereview.chromium.org/7064005/ is a NULL CL to
double-check whether this CL breaks anything additional or not.

[cpu_(ooo_6.6-7.5), 2011-05-24 17:55:39]

RSLGTM

Let me craft an email to the secteam as a heads up.

On 2011/05/23 22:22:41, shess wrote:
> Apologies.
> 
> I ran things through the trybots earlier, on CL at
> http://codereview.chromium.org/7042020/ .  When I went to re-upload, it ...
> didn't work, or something, lots of odd stuff was going on with completely
> unrelated files.  So I spun off a new CL.  So it's possible I won't be able to
> easily upload changes (though I'll keep my git repo static for now).
> 
> The 3.7.6.2 CL had only a small WAL fix added for 3.7.6.3, so it shouldn't
> impact the eventual results of the current try jobs.
> 
> For future reference, http://codereview.chromium.org/7064005/ is a NULL CL to
> double-check whether this CL breaks anything additional or not.