Re: Virtual generated columns

On 08.08.24 20:22, Dean Rasheed wrote:
> Looking at the rewriter changes, it occurred to me that it could
> perhaps be done more simply using ReplaceVarsFromTargetList() for each
> RTE with virtual generated columns. That function already has the
> required wholerow handling code, so there'd be less code duplication.

Hmm, I don't quite see how ReplaceVarsFromTargetList() could be used
here. It does have the wholerow logic that we need somehow, but other
than that it seems to target something different?

> I think it might be better to do this from within fireRIRrules(), just
> after RLS policies are applied, so it wouldn't need to worry about
> CTEs and sublink subqueries. That would also make the
> hasGeneratedVirtual flags unnecessary, since we'd already only be
> doing the extra work for tables with virtual generated columns. That
> would eliminate possible bugs caused by failing to set those flags.

Yes, ideally, we'd piggy-back this into fireRIRrules(). One thing I'm
missing is that if you're descending into subqueries, there is no link
to the upper levels' range tables, which we need to lookup the
pg_attribute entries of column referencing Vars. That's why there is
this whole custom walk with its own context data. Maybe there is a way
to do this already that I missed?