Hi, I would like to use SQLite in a scripting context where there are thousands of "potential" virtual tables, but of these, only a very very small subset (a couple or a handful at most) would be actually used in any given script, so it wouldn't make sense to expose them all, all the time.
I'm looking for a way to leverage the existing SQLite SQL parser to figure out which of those tables are actually used, and expose only those. Basically, some approach like - pass the SQL without exposing anything, which could result in an error - if the error correspond to "unknown table", expose the table - repeat until there are no more errors (or a "real" error is encountered) I could use a custom SQL parser, but that would be complicated, would have to be kept compatible with SQLite evolutions, and the parsing would be completely wasted when the tables have already been exposed. Parsing the error message is simple and works, but it's also fragile if the message text is changed... Is there a more reliable way? Something like an sqlite3_errentity() that would return the unknown table name when the sqlite3_errcode() says the error is an unknown table, that would return the unknown column name when the errcode is an unknown field etc. Eric _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
