Re: [sqlite] _ROWID_ internal column, versus explicit INTEGER PRIMARY KEY with VACUUM

Andrew Finkenstadt [EMAIL PROTECTED] wrote: I realize that FTS1/2 has this slight flaw with the text indexes recording the _rowid_ of a table, in the expectation that a rowid was permanent. That would have caught me unawares, as in Oracle a ROWID is permanent... even if the row has migrated,

[sqlite] _ROWID_ internal column, versus explicit INTEGER PRIMARY KEY with VACUUM

I realize that FTS1/2 has this slight flaw with the text indexes recording the _rowid_ of a table, in the expectation that a rowid was permanent. That would have caught me unawares, as in Oracle a ROWID is permanent... even if the row has migrated, there's a migrate record at the place where the