Re: display temp table structure?

Bruce Momjian wrote:

> > Tom Lane wrote:
> > >
> > > Bruce Momjian <pgman@candle.pha.pa.us> writes:
> > > > There is no user-visible mapping, though the number is the process id.
> > > > You could use that.  I think you can get the backend process id somehow,
> > > > but I am not sure how.  Anyone?

It seems that client application has to keep a mapping for all the temporary
tables it created.  I wonder if that's the case, how can you later implement
\d temp_table_name  (in the TODO list)?

If backend does not keep temp_table_name, this looks like somthing impossible to
do (to me).  (I hope my question makes sense.)

BTW, thanks for the workaround.


>
> > > If you're using libpq, there is a function to retrieve the backend's PID
> > > from the PGconn (PQbackendPID).  I don't think any of the other client
> > > libraries provide this (yet).
> >
> > I installed getpid() as a user-defined function (while testing an idea)
> > - it seemed to work and I assumed it was the backend pid I was getting.
>
> Yes, that is the _nify_ fix I was thinking about.  Use the proper path
> for libc and:
>
>   CREATE FUNCTION getpid () returns int4 as '/lib/libc.so.6' LANGUAGE 'C';
>   SELECT getpid();
>

--
LM.Liu