Bug 68465 – Do not consider SQLITE_BUSY an error in turnOnIncrementalAutoVacuum

• Component: WebCore Misc.
• Status: RESOLVED WONTFIX

[Raphael Kubo da Costa (:rakuco)]

Do not consider SQLITE_BUSY an error in turnOnIncrementalAutoVacuum

[Raphael Kubo da Costa (:rakuco)]

Created attachment 108049 [details]
Patch

[Brent Fulgham]

Comment on attachment 108049 [details]
Patch

Is it possible that this code was actually meant to be "if (error != SQLITE_BUSY)" rather than the original "if (error != SQLITE_ROW)"?

I don't see anything in the SQLite documentation that mentions that SQLITE_ROW is an expected error result of PRAGMA auto_vacuum.

[Raphael Kubo da Costa (:rakuco)]

Perhaps SQLITE_ROW can come from some call invoked in SQLiteStatement::getColumnInt? I'm not sure either.

[Michael Nordman]

In the context of WebSQLDatabase, this change looks harmless. Afaict the net effect is to avoid making a LOG_ERROR() call when the error is BUSY. I don't see any other callsites in the codebase to this method other than the one the AbstractDatabase class.

But a return value of true in this case might be misleading, the auto_vacuum setting may or may not be set to the incremental mode. Another way to avoid the LOG_ERROR() output could be to test lastError() for the BUSY value after a false return at the callsite in AbstractDatabase.

I'm not a reviewer here, so i can't help you out any further than that :)

[Anders Carlsson]

Comment on attachment 108049 [details]
Patch

Clearing review flag on patches from before 2014. If this patch is still relevant, please reset the r? flag.

[Anne van Kesteren]

This is a feature that's disabled/removed.