Ed Leafe wrote: > On May 11, 2009, at 11:42 AM, Paul McNett wrote: > >> I just wanted to highlight the fact that this commit is potentially >> dangerous, as I >> know you are aware. I don't understand, at first glance at least, >> why you want >> _fieldStructure to be the same dict for all cursors descended from >> dCursorMixin. >> Doesn't it need to contain the field structure for *this* cursor and >> not *all* cursors? > > Have you looked at the keys for this dict? Can you find a case where > a problem could actually occur?
It's supposed to contain keys on a field name mapped to a value of a field type. It exists so we don't need to look it up each time. But what if you have 2 tables with a 'pkid' field and in one table the type is char and the other table the type is int? Wouldn't the cursor that was instantiated last get the field type as set from the first cursor? All I wanted to point out is that this smells very dangerous, assigning a mutable object to a class attribute. I get the same smell when I see a mutable object as a function default value. It seems you are trying to make the child cursors of the parent bizobjs not need to constantly call getFields(), which would be a good thing to circumvent. Since the concept of child tables is completely defined in dBizobj, and not in dCursorMixin, perhaps the right place for this optimization is at that layer instead. Paul _______________________________________________ Post Messages to: [email protected] Subscription Maintenance: http://leafe.com/mailman/listinfo/dabo-dev Searchable Archives: http://leafe.com/archives/search/dabo-dev This message: http://leafe.com/archives/byMID/[email protected]
