Richard Hipp <[email protected]> writes: > On Tue, May 17, 2011 at 5:11 PM, Nikolaus Rath > <[email protected]> wrote: >> Please consider the attached testcase. The WAL file should grow to at >> most 1 MB. This limit is exceeded, because the script adds data while a >> second cursor has an active SELECT query. However, when the SELECT query >> finishes, the WAL file is not auto-checkpointed either. >> >> An access pattern similar to the one in the testcase thus results in >> unbounded growth of the WAL file, even though auto-checkpointing is >> enabled, and there are time slots where it could be performed (no >> transaction active). >> >> $ python bug_or_not.py test.sqlite >> Max WAL file should be 1 MB >> Starting query with cursor 2.. >> Inserting 5 MB of data with cursor 1 >> Size of WAL file is now: 16467.0 kB >> Finishing cursor 2 query.. >> Size of WAL file is now: 16467.0 kB >> Starting query with cursor 2.. >> Inserting 5 MB of data with cursor 1 >> Size of WAL file is now: 32954.0 kB >> Finishing cursor 2 query.. >> Size of WAL file is now: 32954.0 kB >> >> Bug or not? >> > > Not. It is working as it was designed and documented to work. Perhaps you > mean that you want to start a discussion about the design, perhaps leading > to a change in design. > > I'm rather pedantic about the use of the word "bug". To me, a "bug" means > it gets the wrong answer. And for the purpose of that definition, crashing > is considered the wrong answer. Failing to do a checkpoint when you think > it ought to, even though it is checkpointing as advertised, is not a "bug". > You can argue that there is a need to enhance auto-checkpoint to make it > more programmer-friendly. But that is a feature request, not a bug.
That's why I posed it as a question, not a statement :-). My apologies if it still came across the wrong way. So considering this as feature request: do you have any thoughts on making SQLite consider auto-checkpointing when a read transaction finishes? Best, -Nikolaus -- »Time flies like an arrow, fruit flies like a Banana.« PGP fingerprint: 5B93 61F8 4EA2 E279 ABF6 02CF A9AD B7F8 AE4E 425C _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
