Dear Horiguchi-san,
Thank you for giving comments!
> Thanks for the new version. I don't object for reusing
> process_log_prefix_padding, but I still find it strange that the
> function with the name 'process_padding' is in string.c. If we move
> it to string.c, I'd like to name it "pg_fast_strtoi" or something and
> change the interface to int pg_fast_strtoi(const char *p, char
> **endptr) that is (roughly) compatible to strtol. What do (both) you
> think about this?
I agree that this interface might be confused.
I changed its name and interface. How do you think?
Actually I cannot distinguish the name is good or not,
but I could not think of anything else...
> I didn't fully checked in what case parse_pgfdw_appname gives "" as
> result, I feel that we should use the original value in that
> case. That is,
>
> > parse_pgfdw_appname(&buf, vaues[i]);
> >
> > /* use the result if any, otherwise use the original string */
> > if (buf.data[0] != 0)
> > values[i] = buf.data;
> >
> > /* break if it results in non-empty string */
> > if (values[0][0] != 0)
> > break;
Hmm. I tested log_line_prefix() by setting it as '%z'
and I found that any prefixes were not output.
I think we should follow it, so currently not fixed.
Best Regards,
Hayato Kuroda
FUJITSU LIMITED