Re: Views "missing" from information_schema.view_table_usage

Oh ok. Not to cause confusion, but after I suggested I would request an
update to the docs, I thought maybe it would be better to ask if the VTU's
code could be modified to include MVs. So I sent a request to
pgsql-hackers (I think that's the list to use for feature requests). But
from what you're saying, Tom, that may be a fruitless endeavor. If they
reply back "nope", then I'll submit a request to have the documentation
updated.

Thanks for everyone's contributions!

Jon

On Fri, Dec 2, 2022 at 5:11 PM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:

> Erik Wienhold <ewie(at)ewie(dot)name> writes:
> > Could it be a bug? Materialized views are a Postgres extension[1] (I
> always
> > thought they are standard.) But I'd expect them to be included when
> talking
> > about "views". Maybe they are not included because they are considered
> being
> > closer to physical tables[2] than views. Yet their dependencies would
> justify
> > inclusion in view_table_usage.
>
> The reasoning is that the information_schema views are defined by the
> SQL standard and therefore should only show content that matches the
> standard. Thus, they ignore PG-invented objects like matviews and
> sequences. Some other projects adopt more liberal views about
> what should be shown in those views, but that one is our policy.
>
> regards, tom lane
>