just tried out your checker. I guess what it really needs is to get the information dynamically so the checker is always uptodate. That being said things like "GROUP" are probably a reserved word in any RDBMS since ages, but does not show up as reserved in all RDBMS in your checker.

BTW: I am the lead author of PEAR::MDB2_Schema, which essentially is a frontend to manage a database schema in an independent format (currently the only format supported is an xml schema format). One of the cool features is that it can check the identifiers against lists of reserved words.

Anyways I will try if I can establish a standard for publishing reserved words lists in the open source database world.

Keywords in the PostgreSQL grammar are divided into several groups, and the so-called "unreserved" keywords are actually available for use as table or column names, or in any other context as any non-keyword is. The grammar needs them as keywords because they're used in some non-ambiguous context, but for an application, there is no disinction between an unreserved keyword and a word that's not a keyword at all. For example, "CHECKPOINT" is an unreserved keyword because there's a CHECKPOINT command to perform a checkpoint, but you can create a table called CHECKPOINT or column with no problem.

An up-to-date list of keywords can be found at http://www.postgresql.org/docs/8.4/interactive/sql-keywords-appendix.html. You should ignore all the keywords listed as "non-reserved" (or blank). Even better, you could distinguish the classes of "reservedness" in the tool, but you should leave out the non-reserved ones anyway.

I plan to add a tool checking a schema for reserved keywords into the Doctrine ORM (www.doctrine-project.org) and having a knowledgeable database as backend is better than maintaining our own one.

You would be mentioned with URL and credits also :)