This is true. It would be cleaner, though, if we could check for an
attribute in pg_class. I do not recall one for that purpose.
Thomas G. Lockhart wrote:
> > I'm running into some naming problems while doing so. Having
> > pg_table, pg_view etc. as views lets a users assume pg_index
> > would be one too where to get some information. But pg_index
> > already exists.
> >
> > Should I name all of them pgv_... ?
> >
> > Other databases have many views starting with DBA or SYS on
> > the other hand. For now I'll start naming them pgv_..., we
> > could rename them before applying the patch.
>
> I recall that there are some places in the code (maybe only in the
> client-side drivers?) which check explicitly for a "pg_%" pattern to
> decide if a table or resource is a system table.
>
> How about "pg_index_v", for example?
>
> - Tom
