Many thanks to all who replied, that was very helpful. So the ticket was closed because the old CVSTRACKER was closed, and a new ticket should be created with the replacement mechanism, after being discussed in the mailing list. @Roger: I apologize for the misunderstanding.
> If > it alters the behaviour compared to previous SQLite versions then breaking > that backwards compatibility would require exceptionally compelling evidence. > In my opinion fixing this error should not break compatibility, because I can't imagine a legitimate usage case that relies on a syntax error in order to work correctly. Breaking compatibility in this case would mean that some old code that didn't work suddenly starts working. Would that be so bad? Also, I was unable to find a practical workaround for this issue. Since in my case the queries are auto-generated, the only thing that comes close would be to parse the output and replace the '*' with the full list of columns in the table. I tried to feed a COCO/R generated parser with the SQLite grammar but it choked and died miserably, so maybe I'll have better luck with lexx/yacc. Personally I am inclined to re-open this ticket with the new bug tracker, unless someone could give a reason why not to. Many thanks. Bogdan _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
