"B V, Phanisekhar" <[EMAIL PROTECTED]> wrote in message news:[EMAIL PROTECTED] > So is it possible > that SQLITE will give an error when attempting to insert a record even > if there are free ROWID's?
Yes, in theory. If you have close to 2^63-1 rows, probing at random for a ROWID to reuse may just happen to keep hitting occupied ROWIDs. If I remember correctly, after a certain number of unsuccessful attemps SQLite gives up and returns an error. Of course, it is highly unlikely that you will ever have anywhere close to 2^63 rows. Storage requirements alone are prohibitive. The database will likely become unusable long before that. In fact, the only realistic way to take the highest possible ROWID, and thus force SQLite into "pick at random" mode, is to explicitly specify the large ROWID when inserting a record. So just don't do that. > Please consider this scenario in the worst case. I agree that the > number > of ROWID's is so large, but still there is a possibility it might be > filled. If you insert a new record every millisecond, it will take some 300 million years to exhaust ROWIDs. What kind of software do you develop that you feel this might come up as even a remote possibility? > I basically want to understand the algorithm used by SQLITE in > picking a new ROWID in such scenario. Which part of "at random" do you find unclear? Igor Tandetnik _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
