Thread: bytea and vacuum analyze on postgresql 7.3.4

Hi,

I hit the following bug:

this is with a 7.3.4 database server (and psql client)
PGCLIENTENCODING=3D"UTF-8"
and a utf-8 encoded database,

pensieve=3D> create table a(raw bytea);
CREATE TABLE
pensieve=3D> insert into a values('\770\711\056'::bytea);
INSERT 1112630 1
pensieve=3D> vacuum analyze a;
VACUUM
pensieve=3D> insert into a values('\770\711\056'::bytea);
INSERT 1112631 1
pensieve=3D> vacuum analyze a;
ERROR:  Invalid UNICODE character sequence found (0xf8c92e)

I think bytea should accept ANY value and vaccuum analyze must
not complain about this.

Is this new to you?

Regards,


Michael


--=20
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D
Michael Radziej           SuSE Linux AG              phone +49-911-74053-0
IT                        Deutschherrenstr. 15-19    fax   +49-911-3206727
(vormals "interne EDV")   90429 N=FCrnberg, Germany    web   http://www.sus=
e.de
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D
 Heute schon ge=E4ndert?                                      -- simply cha=
nge

Michael Radziej <mir@suse.de> writes:
> I think bytea should accept ANY value and vaccuum analyze must
> not complain about this.

This is fixed as of 7.4, I think (at least your example case doesn't
fail for me).  7.3 was a bit sloppy about how it treated entries in
pg_statistic ...

            regards, tom lane