[sqlite] Allow recover.c to compile independently.

[sqlite] Allow recover.c to compile independently.

Last step to removing recover.c's dependency on SQLite internals.

recover_varint.c copies the getVarint() implementation from
fts5_varint.c, and renames to recoverGetVarint().  Copied from fts5
because I hope to upstream this to SQLite's ext/ directory someday.

In recover.c, cribbed some typedefs from fts5Int.h.  Replaced Pgno with
u32 (pager.h defines it as u32).  Renamed getVarint() calls.  Virtual
table creation parser assumes "main" db rather than "name of first
db" (they should be the same).

At this point, recover.c can be compiled outside the amalgamation, but
doing so would break Chromium's build.

BUG=584407
Committed: https://crrev.com/481e5587ab14da2fcea690288421499d762ed32f
Cr-Commit-Position: refs/heads/master@{#379902}

[Scott Hess - ex-Googler, 2016-03-01 23:41:21]

The CQ bit was checked by shess@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-03-01 23:41:57]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1753993002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1753993002/1

[Scott Hess - ex-Googler, 2016-03-02 01:12:08]

Annotate the various changes.

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/amalgama...
File third_party/sqlite/amalgamation/sqlite3.c (right):

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/amalgama...
third_party/sqlite/amalgamation/sqlite3.c:136714: +/* #include <stdint.h> */
This file is generated by a script from files in sqlite/src/src, so not
important to review.

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/Make...
File third_party/sqlite/src/Makefile.in (right):

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/Make...
third_party/sqlite/src/Makefile.in:264: $(TOP)/src/recover_varint.c \
This file is used to build SQLite's internal test suite, which I've modified to
include recover tests.

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/Make...
File third_party/sqlite/src/Makefile.linux-gcc (left):

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/Make...
third_party/sqlite/src/Makefile.linux-gcc:87: 
Was only needed to testfixture, moved to main.mk.

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/src/...
File third_party/sqlite/src/src/recover.c (right):

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/src/...
third_party/sqlite/src/src/recover.c:222: typedef uint32_t u32;
The versions in fts5int.h were unsigned char and unsigned int, which made me
somewhat nervous.

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/src/...
third_party/sqlite/src/src/recover.c:427: u32 pgno;                      /* Page
number for this page */
pager.h has: typedef u32 Pgno;.  Plain unsigned would also work, it just needs
to be 32-bit or larger.

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/src/...
third_party/sqlite/src/src/recover.c:2236: pRecover->zDb =
sqlite3_strdup("main");
aDb[0] is "main", originally this code poked into the sqlite3 and other structs
quite a bit.

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/src/...
File third_party/sqlite/src/src/recover_varint.c (right):

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/src/...
third_party/sqlite/src/src/recover_varint.c:9: **    May you share freely, never
taking more than you give.
I'm hoping/intending to upstream this, so it has the SQLite boilerplate.

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/src/...
third_party/sqlite/src/src/recover_varint.c:13: ** Copy of
sqlite3Fts5GetVarint() from fts3_varint.c, which in turn is copied
fts5_varint.c, rather.

fts5_varint.c is the file on the left.  All of the deleted stuff is stuff
recover.c doesn't need, so the gist of it is "a renamed function, plus a bit of
header copied in".

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/src/...
third_party/sqlite/src/src/recover_varint.c:23: typedef unsigned int   u32;
fts5Int.h, rather.  Didn't mess with these, because the code being copied is all
SQLite code, so I'm assuming they know what they're doing.

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/tool...
File third_party/sqlite/src/tool/mksqlite3c.tcl (right):

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/tool...
third_party/sqlite/src/tool/mksqlite3c.tcl:365: recover_varint.c
The overall script builds third_party/sqlite/amalgamation/sqlite3.c .

[commit-bot: I haz the power, 2016-03-02 01:13:03]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-03-02 01:13:04]

Dry run: Try jobs failed on following builders:
  linux_chromium_rel_ng on tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[Scott Hess - ex-Googler, 2016-03-02 18:45:19]

shess@chromium.org changed reviewers:
+ sdefresne@chromium.org

[Scott Hess - ex-Googler, 2016-03-02 18:45:20]

Thanks for volunteering!

[sdefresne, 2016-03-08 15:42:37]

lgtm

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/src/...
File third_party/sqlite/src/src/recover_varint.c (right):

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/src/...
third_party/sqlite/src/src/recover_varint.c:28: ** Bitmasks used by
sqlite3GetVarint().  These precomputed constants
nit: should this "sqlite3GetVarint()" be changed to "recoverGetVarint()" too?

[Scott Hess - ex-Googler, 2016-03-08 18:37:36]

Thanks!

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/src/...
File third_party/sqlite/src/src/recover_varint.c (right):

https://codereview.chromium.org/1753993002/diff/1/third_party/sqlite/src/src/...
third_party/sqlite/src/src/recover_varint.c:28: ** Bitmasks used by
sqlite3GetVarint().  These precomputed constants
On 2016/03/08 15:42:36, sdefresne wrote:
> nit: should this "sqlite3GetVarint()" be changed to "recoverGetVarint()" too?

Maybe?  The SQLite people didn't adapt it for fts5.  I'll change it, though,
just to remove the non sequitur.

[Scott Hess - ex-Googler, 2016-03-08 18:38:17]

Comment change

[Scott Hess - ex-Googler, 2016-03-08 18:38:44]

The CQ bit was checked by shess@chromium.org

[Scott Hess - ex-Googler, 2016-03-08 18:38:44]

The patchset sent to the CQ was uploaded after l-g-t-m from
sdefresne@chromium.org
Link to the patchset: https://codereview.chromium.org/1753993002/#ps20001
(title: "Comment change")

[commit-bot: I haz the power, 2016-03-08 18:39:13]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1753993002/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1753993002/20001

[commit-bot: I haz the power, 2016-03-08 20:41:29]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-03-08 20:42:56]

Description was changed from

==========
[sqlite] Allow recover.c to compile independently.

Last step to removing recover.c's dependency on SQLite internals.

recover_varint.c copies the getVarint() implementation from
fts5_varint.c, and renames to recoverGetVarint().  Copied from fts5
because I hope to upstream this to SQLite's ext/ directory someday.

In recover.c, cribbed some typedefs from fts5Int.h.  Replaced Pgno with
u32 (pager.h defines it as u32).  Renamed getVarint() calls.  Virtual
table creation parser assumes "main" db rather than "name of first
db" (they should be the same).

At this point, recover.c can be compiled outside the amalgamation, but
doing so would break Chromium's build.

BUG=584407
==========

to

==========
[sqlite] Allow recover.c to compile independently.

Last step to removing recover.c's dependency on SQLite internals.

recover_varint.c copies the getVarint() implementation from
fts5_varint.c, and renames to recoverGetVarint().  Copied from fts5
because I hope to upstream this to SQLite's ext/ directory someday.

In recover.c, cribbed some typedefs from fts5Int.h.  Replaced Pgno with
u32 (pager.h defines it as u32).  Renamed getVarint() calls.  Virtual
table creation parser assumes "main" db rather than "name of first
db" (they should be the same).

At this point, recover.c can be compiled outside the amalgamation, but
doing so would break Chromium's build.

BUG=584407

Committed: https://crrev.com/481e5587ab14da2fcea690288421499d762ed32f
Cr-Commit-Position: refs/heads/master@{#379902}
==========

[commit-bot: I haz the power, 2016-03-08 20:42:57]

Patchset 2 (id:??) landed as
https://crrev.com/481e5587ab14da2fcea690288421499d762ed32f
Cr-Commit-Position: refs/heads/master@{#379902}

[evgeny.s.sidorenko, 2016-06-16 13:06:29]

Hi Scott,
it looks like you forgot to add your patch to third_party/sqlite/patches/ as
mentioned in README.chromium file.

[Scott Hess - ex-Googler, 2016-06-16 23:38:41]

On 2016/06/16 13:06:29, evgeny.s.sidorenko wrote:
> it looks like you forgot to add your patch to third_party/sqlite/patches/ as
> mentioned in README.chromium file.

OK, I'll go look at that.  I may have forgotten to land a final CL to move the
module from sqlite/src/src to something Chromium-local like sqlite/ext/ .