> Below were the communications between Tom and me before I implemented this
> project.  I just did what he asked me to do.

Part of it, maybe --- my point was that without any support in (at
least) plpgsql, the feature is of only academic interest.  There's not
a lot of point in applying the patch when it does not do anything.

Also, we tend to look with suspicion on such stuff because once you
actually write code that uses the feature, you often find that you
should have designed it a little differently.  Nailing down the catalog
representation in advance of having working code that does something
useful with it is a good recipe for making mistakes.  (To take one
example, why does the patch only support one name?  Don't you need two
for the UPDATE case?)

In any case the patch is missing documentation and pg_dump support,
making it even less possible to use it for anything.  It's project
policy that all system catalog columns be documented in catalogs.sgml,
and what's the use of DDL that won't survive a dump and reload?

                        regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
      joining column's datatypes do not match

Reply via email to