Thank you Dan, as it is well behind my understanding of Digikam, I've asked one of the lead developper to have a look upon your answer and its implications.
Best regards, -Mathieu 2014-07-15 22:27 GMT+02:00 Dan Kennedy <danielk1...@gmail.com>: > On 07/16/2014 03:22 AM, Dan Kennedy wrote: > >> On 07/15/2014 09:06 PM, Mathieu Clabaut wrote: >> >>> Hello, >>> >>> Digikam bug https://bugs.kde.org/show_bug.cgi?id=323888#c89 seems to >>> be >>> caused by a sqlite memory leak >>> >> >> It looks very much like the program is not calling sqlite3_close(). >> >> If you have a small program that shows SQLite leaking memory please make >> it available to us. >> >> >> and is said to be corrected with slqlite >>> 3.8.5, but as shown in https://bugs.kde.org/show_bug.cgi?id=321680 a >>> similar problem appear in digikam 4.0.0 with sqlite 3.8.5. >>> >> >> I'm not sure what causes the crash in the stack trace in comment 8. I >> don't think these crashes are related to bug 323888 in any case. It's tough >> to say what the trace in comment 8 means really, except that the heap is >> corrupted. >> >> The crashes in sqlite3MemCompare() are interesting though. Does the >> framework ever pass anything other than SQLITE_TRANSIENT or SQLITE_STATIC >> to sqlite3_bind_text(), bind_result() or similar? >> > > Actually - it could be that a bad function pointer has been passed to one > of the sqlite3_create_collation() functions. That would give you very > similar stack traces anyhow. > > > Dan. > > > _______________________________________________ > sqlite-users mailing list > sqlite-users@sqlite.org > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
