On Sun, Aug 13, 2006 at 08:51:53PM -0600, Richard Greenwood wrote: > The 30 char versus 60 char issue still sounds like encoding: I think > MapInfo is interpreting 2 byte encoding as 1 byte encoding. So a two > byte char type appears to MapInfo to be twice as wide as it actually > is. UTF8 is 1 byte (8 bit) encoding. Not sure what LATIN1 is. I think > UNICODE is 2 byte.
It does sound suspiciously like an encoding problem, but setting the cluster and database's encoding to LATIN1 (ISO-8859-1, which is a 1-byte system) should have done the trick, but it didn't. And according to the PostgreSQL documentation's section 21.2 "Character Set Support", UTF8 is actually a multi-byte system supporting up to 4 bytes per character, and its alias name is 'Unicode'. Perhaps it has to do with how my Fedora Core 4 Linux is handling the encoding in its environment. It's still a mystery, and even more so because I can modify these fields with MapBasic and MS Access VBA with no problems. Only MapInfo's interactive interface seems to have a problem with it. > Are all fields in the TAB file readonly, or just the char field? I > have often seen that a primary (unique) key is required for an linked > ODBC table to be updateable. But I would think that would effect all > fields in the table. Ah, yes... It's just that character field and the OIDs field. All the other fields are numeric. And you're right about the index. You do have to have a unique index set before an ODBC table will be read/write, but I did that with the numeric 'id' field. _______________________________________________ MapInfo-L mailing list [email protected] http://www.directionsmag.com/mailman/listinfo/mapinfo-l
