Hi,
My application's architecture looks roughly like this:
- A main thread which does only database reads
- A dedicated database write thread
I needed to create this because while reads are fast, there are
occasions where writes take a significant portion of time and that can
cause the main thread to lock up, which results in a sub-par user
experience.
The problem I am having is that somewhere in the main thread, something
is grabbing a shared lock on the SQLite DB and not letting go. the
effect of this is that the dedicated write thread ends up waiting until
the application exits(!) in order to do some writes that occur after this.
I've tried debugging manually, disabling various features but I haven't
been able to pinpoint which bit actually grabs the lock. Are there any
debugging strategies I can employ to get to the bottom of it?
My current journal_mode is DELETE but WAL has similar problems, in that
the database writes are not seen, presumably because some reader has an
open transaction somewhere.
Thanks for your help.
Sohail
_______________________________________________
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
