On Jun 2, 2010, at 7:19 PM, Simon Slavin wrote: > > On 2 Jun 2010, at 10:12am, Dan Kennedy wrote: > >> The fix you propose would ignore the (suspected) corruption >> and continue without reporting it to the user. Which might be >> the best thing for some > > but not for me. Please do not roll anything like that into the > source. I definitely want error messages if there's even the > slightest hint that my database is corrupt. > > What I would like to know is why the PRAGMA check shows no > corruption in that database. Perhaps it could be revised to do so.
None of the database tables or indexes are corrupt. So "PRAGMA integrity_check" sees no problems. It appears that the content of the various tables used by the rtree module to store the content of a single R-Tree structure is inconsistent though. The database is corrupt at the application level, if you like. The application being the rtree virtual table module. Could be there is a bug in the rtree module causing this. Could be the underlying tables were modified externally, bypassing rtree. You could use a foreign key constraint to stop this happening (some cases, anyway). You probably wouldn't, because it would slow down the rtree module, but you could. (Hey, an actual use for foreign keys!) _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
