Chris writes:
> I'm resubmitting this patch from a while ago, now that 7.0 is out.
I don't recall that discussion ever ending with a definite design
document. Could you at least point out which of the many proposals this
refers to? What I do remember in fact is that the core decided that the
various proposals need to be checked against SQL3 before anything can
happen and as far as I can see you definitely didn't do that.
> If you cast your minds back, this patch allows update and delete to
> work on inheritance hierarchies just like it now works on select.
I don't think that's a good idea. If I have an inheritance hierarchy A < B
< C and I update a row in A "only" then I break the hierarchy. (I'm also
wondering how you would do that, since you would have to make a copy of
the row for A and then keep the old copies around for B and C.) SQL3
violation right there. That also goes for the various ALTER TABLE [ONLY]
syntax additions. If I add a row to A only then B is no longer a subtable
of A. One thing I see you didn't change is the CREATE TABLE syntax,
although that could posibly have used it.
> It also uses the Informix/Illustra model for subclasses - i.e. "ONLY",
> as was discussed at length before.
SELECT ONLY is cool, that's the SQL3 way.
All in all I think it's great that you're tackling this but this patch
looks very suspect to me. Rolling your own inheritance model when there's
decades of scientific research out there that some presumably smart guys
wrote down in a (now official) standards document doesn't seem like a wise
thing to do.
PS: Could you elaborate on that FAQ_DEV change?
--
Peter Eisentraut Sernanders väg 10:115
peter_e@gmx.net 75262 Uppsala
http://yi.org/peter-e/ Sweden
