> > usa=# select ''::integer;
> > ERROR: pg_atoi: zero-length string
>
> > australia=# select ''::integer;
> > WARNING: pg_atoi: zero-length string
> > int4
> > ------
> > 0
> > (1 row)
> >
> > What the???
>
> Looks like the FreeBSD maintainer (you know who you are, seanc :-) )
> has decided to be clever:
>
>
http://www.freebsd.org/cgi/cvsweb.cgi/ports/databases/postgresql7/files/patch-src%3a%3abackend%3a%3autils%3a%3aadt%3a%3anumutils.c?rev=1.1&content-type=text/x-cvsweb-markup
I don't think it was clever so much as bowing to the whining masses of
RT and Horde users and I caved...
> I had asked that it be ****OPTIONAL****, and Sean had ignored that
> request.
LER, you were only one of thirty people that nagged me on this, but
you were the one who submitted the patch and were the 1st to point it
out. :)
> > australia=# select ''::integer;
> > WARNING: pg_atoi: zero-length string
>
> That's got to be a locally-modified copy. The "zero-length string"
> text has only existed since last August, and it's always been an ERROR.
Hrm, well, if I don't get any criticism or a heads up for when the
7.3.3 release is (hint hint) in the next 48hrs, I'll dawn my flame
retardant suite and hope that the Horde/RT folks have a less broken
product at this time. I'll remove the patch and bump the port version
for those interested.
> Yes, I agree the distro should not modify this, but I do think it
> would have been better to have it as a WARNING at least for one
> release...
One of the goals of the FreeBSD ports system is to mitigate these
inconsistencies in programs/packages and to help provide an easier to
use platform that is more controlled than native programs found out in
the wild. These kinds of patches are pretty common, but it is rare
that PostgreSQL has 'em because of it's releng process is
substantially better than other projects out there. -sc
--
Sean Chittenden