|
From: Twylite <tw...@cr...> - 2008-06-16 20:04:17
|
> > From: rna020 <tom...@fr...> > > This can be handled by the simple suggestions being proposed by the "no > NULL in tcl" side of the discussion. What isn't clear, at least to me, > is weather these ideas scale to more complicated cases in SQL and other > situations like holes in lists. > That really depends on the people specifying the APIs for those subsystems. If the "nextrow" in TDBC returned a list of tuples where each tuple was "column-name type value" then we wouldn't be having this discussion about NULLs because they could be indicated in the type field of the tuple. On the other hand we _would_ be having discussions about how to easily and efficiently retrieve individual values and entire rows from this list-of-tuples, which is why I haven't proposed it seriously ;) Any API that deals with NULLs on the C side is going to have to translate the concept intelligently in a domain-specific manner to Tcl. The file/chan, Tk, and other APIs manage to do this gracefully. SQL is somewhat special in that there are times when you want to know _in Tcl code_ whether the DBMS / SQL query returned a NULL or not. Regards, Twylite |