Re: [HACKERS] pg_dump, problem with user defined types?

> > Looks like I am going to need some help here.
> > The old code dumped out regproc fields as the pg_proc.proname.
> > There is a problem with this. First, you can have multiple proname
> > entries with the same proname. The differ in their argument
> > number/types. The old code, when reading in a regproc name, would do
> > a sequential scan of the pg_proc table, and find the first entry that
> > matches the given proname.
> > If that is not the one you wanted, too bad. No way to change it.
>
> Hi Bruce. I'm sorry again for being so slow, but I'm still not
> understanding the initial conditions which prompted these changes. Are
> you fixing something proactively, or was there a specific example of
> misbehavior? The example I see in your mail with Tatsuo which now causes
> trouble is for type input and output routine names, which _are_ likely
> to be unique.
>
> Would it be possible for you to bracket the code in the cvs tree so that
> we can enable/disable the old behavior? That way we can see what has
> changed and how it used to behave. I suppose that would involve
> bracketing code in regprocin/out and in pg_dump??

Proactive fix. See my other posting today that has an idea to roll back
the old behavour, while making sure the regproc name is unique.

--
Bruce Momjian | http://www.op.net/~candle
maillist(at)candle(dot)pha(dot)pa(dot)us | (610) 853-3000
+ If your life is a hard drive, | 830 Blythe Avenue
+ Christ can be your backup. | Drexel Hill, Pennsylvania 19026