Closed
Bug 1162643
(SQLite3.8.10.1)
Opened 9 years ago
Closed 9 years ago
Upgrade to SQLite 3.8.10.1
Categories
(Toolkit :: Storage, defect)
Toolkit
Storage
Tracking
()
RESOLVED
FIXED
mozilla41
Tracking | Status | |
---|---|---|
firefox41 | --- | fixed |
People
(Reporter: RyanVM, Assigned: RyanVM)
References
()
Details
Attachments
(1 file, 2 obsolete files)
425.70 KB,
patch
|
mak
:
review+
|
Details | Diff | Splinter Review |
SQLite Release 3.8.10 On 2015-05-07 * Added the sqldiff.exe utility program for computing the differences between two SQLite database files. * Added the y format string to the matchinfo() function of FTS3. * Performance improvements for ORDER BY, VACUUM, CREATE INDEX, PRAGMA integrity_check, and PRAGMA quick_check. * Fix many obscure problems discovered while SQL fuzzing. * Identify all methods for important objects in the interface documentation. (example) * Made the American Fuzzy Lop fuzzer a standard part of SQLite's testing strategy. * Add the ".binary" and ".limits" commands to the command-line shell. * Make the "dbstat" virtual table part of standard builds when compiled with the SQLITE_ENABLE_DBSTAT_VTAB option. Richard, we recently upgraded to 3.8.9 across a number of older releases in order to pick up the fuzzing fixes it contained. Is it worth taking this release across all of them too?
Flags: needinfo?(drh)
Comment 1•9 years ago
|
||
The 3.8.10 does fix several arcane bugs (mostly assertion faults) that the AFL fuzzer picked up. But none of this are important to FF, I don't believe. On the other hand, the reason we released 3.8.10 early in the first place is to provide the SQLITE_ENABLE_DBSTAT_VTAB compile-time option, for the DBSTAT virtual table, which Ben Turner wants to use in order to derive better estimates on when a VACUUM of the IndexedDB data would be helpful. Also, the performance improvements result in about 7.5% fewer CPU cycles for our tests.
Flags: needinfo?(drh)
Assignee | ||
Comment 2•9 years ago
|
||
OK, sounds like we can just land it on mozilla-central and let it go through the regular release process then. Thanks!
And we need to add the SQLITE_ENABLE_DBSTAT_VTAB define to our build config.
Assignee | ||
Comment 4•9 years ago
|
||
https://treeherder.mozilla.org/#/jobs?repo=try&revision=a4aa82c4df27
Comment on attachment 8602878 [details] [diff] [review] Upgrade to SQLite 3.8.10 This will need another sqlite3_compileoption_used check in configure.in
Comment 6•9 years ago
|
||
Ugh. In the rush to get the release out, we failed to add SQLITE_ENABLE_DBSTAT_VTAB to the sqlite3_compileoption_used() method. Bummer. Does that throw a wrench into things?
(In reply to D. Richard Hipp from comment #6) > Ugh. In the rush to get the release out, we failed to add > SQLITE_ENABLE_DBSTAT_VTAB to the sqlite3_compileoption_used() method. > Bummer. Does that throw a wrench into things? Oh, hm. Well, I guess we can test for it some other way?
Comment 8•9 years ago
|
||
You mean test at run-time? Just run: "CREATE VIRTUAL TABLE temp.dbstat USING dbstat;" If it works, then SQLITE_ENABLE_DBSTAT_VTAB is on. If it fails, then it isn't.
(In reply to D. Richard Hipp from comment #8) > You mean test at run-time? Well, sorta. We need to test it at configure time (that's half run-time and half build-time!). We need to make sure that any system-supplied sqlite was built with SQLITE_ENABLE_DBSTAT_VTAB, otherwise we make the build fail.
Attachment #8602878 -
Attachment is obsolete: true
Comment 12•9 years ago
|
||
would it be a big deal to release a 3.8.10.1 with updated sqlite3_compileoption_used? Fwiw, we could even take a patched 3.8.10, if the same fix will be in the next Sqlite release. We don't usually patch Sqlite, but it's ok to do that when there's already a fix upstream that is riding the train to the next release.
Comment 13•9 years ago
|
||
oh wait, that wouldn't do for system sqlite, nevermind.
Mak, the updated configure.in changes I attached allow us to detect this without sqlite3_compileoption_used support, so we don't really need a new version, right?
Flags: needinfo?(mak77)
Comment 15•9 years ago
|
||
I saw that, it is still far more complex than it would be with proper sqlite3_compileoption_used support. So I'm asking what's the price to have that fixed in a .1 version, to compare that with the price of this more complex check that regardless we should replace as soon as we get a fixed sqlite3_compileoption_used. If it's easy to get .1 version I don't think it's worth the workaround and fullow-up bug to remove it.
Flags: needinfo?(mak77)
Comment 16•9 years ago
|
||
SQLite version 3.8.10.1 is a patch release that fixes sqlite3_compileoption_used() so that it reports on SQLITE_ENABLE_DBSTAT_VTAB as it should.
(In reply to D. Richard Hipp from comment #16) Thanks!
Attachment #8602993 -
Attachment is obsolete: true
Blocks: 858680
Updated•9 years ago
|
Alias: SQLite3.8.10 → SQLite3.8.10.1
Summary: Upgrade to SQLite 3.8.10 → Upgrade to SQLite 3.8.10.1
Attachment #8604193 -
Flags: review?(mak77)
Comment 19•9 years ago
|
||
Comment on attachment 8604193 [details] [diff] [review] Upgrade to SQLite 3.8.10.1 Review of attachment 8604193 [details] [diff] [review]: ----------------------------------------------------------------- Thanks!
Attachment #8604193 -
Flags: review?(mak77) → review+
Comment 21•9 years ago
|
||
https://hg.mozilla.org/mozilla-central/rev/c6ea21c091e9
Status: NEW → RESOLVED
Closed: 9 years ago
status-firefox41:
--- → fixed
Resolution: --- → FIXED
Target Milestone: --- → mozilla41
Depends on: SQLite3.8.10.2
Blocks: SQLite3.8.10.2
No longer depends on: SQLite3.8.10.2
Comment 22•9 years ago
|
||
SQLite 3.8.10.2 was released on the 5/20/2015 with the below fix being the only update: - Fix an index corruption issue introduced by version 3.8.7. An index with a TEXT key can be corrupted by an INSERT into the corresponding table if the table has two nested triggers that convert the key value to INTEGER and back to TEXT again. Ticket 34cd55d68e0
Comment 23•9 years ago
|
||
(In reply to Arthur K. from comment #22) > SQLite 3.8.10.2 was released on the 5/20/2015 with the below fix being the > only update: > > - Fix an index corruption issue introduced by version 3.8.7. An index with a > TEXT key can be corrupted by an INSERT into the corresponding table if the > table has two nested triggers that convert the key value to INTEGER and back > to TEXT again. Ticket 34cd55d68e0 Ugh, sorry for the churn. Saw too late that this was already reported.
Assignee | ||
Updated•9 years ago
|
No longer blocks: SQLite3.8.9
Depends on: SQLite3.8.9
Assignee | ||
Updated•9 years ago
|
Blocks: SQLite3.8.11.1
You need to log in
before you can comment on or make changes to this bug.
Description
•