Tom Lane wrote:
> Bruce Momjian <pgman@candle.pha.pa.us> writes:
> > I looked over the patch and it seems to continue on pg_restore errors by
> > default. That isn't good. By default, any error should make it exit
> > loudly.
>
> I'm not sure of that. pg_dump is really designed and tested for the
> case of text dump to a psql script, and if there is an error in the psql
> script, psql will print it out and continue. In the cases I have seen
> where pg_restore operates differently from that paradigm, the pg_restore
> behavior of curling up and dying has *never* been the right thing.
>
> I haven't looked at the details of Fabien's patch, but I'm in agreement
> with the overall goal.
Oh, OK, so make it behave like pg_dump's text output piped into psql.
The patch is quite clean and just changes die to warn for pg_restore,
and reports the number of errors on completion.
The only thing that it needs is to change pg_restore so it returns a
non-zero error if there were any warnings printed. I can make that
change easily.
I will add it to the patch queue.
--
Bruce Momjian | http://candle.pha.pa.us
pgman@candle.pha.pa.us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073