The following bug has been logged online:

Bug reference: 1248
Logged by: Timothy LoGrasso

Email address: timlograsso (AT) yahoo (DOT) com

PostgreSQL version: 7.3.5

Operating system: SunOS 5.9

Description: pg_dump produces grants with concatonated elements

Details:

pg_dump produces the following grants:
GRANT INSERTSELECTUPDATE ON TABLE camp_score TO mn;

Obviously there are no commas or spaces seperating the different privs. I'm
actually running the following version: PostgreSQL 7.3.6 on
i386-pc-solaris2.8, compiled by cc -fast -xarch=386 -Xa

Attached is the ddl to create the table and grant permissions.

CREATE USER mn PASSWORD 'mn';
CREATE USER mnview PASSWORD 'mnview';
CREATE USER mgr PASSWORD 'mgr';
CREATE USER mndb PASSWORD 'mndb';

CREATE TABLE camp_score (
ipr_number character varying(15) NOT NULL,
performed_by character(6),
score_date date,
hs_comp smallint,
hs_tech smallint,
hs_hyg smallint,
hs_safe smallint,
hs_fire smallint,
hs_hp smallint,
hs_crit smallint,
hs_infr smallint,
ew_comp smallint,
ew_tech smallint,
ew_liq smallint,
ew_sol smallint,
ew_air smallint,
ew_min smallint,
ew_rest smallint,
ew_corr smallint,
ew_infr smallint,
ss_comp smallint,
ss_tech smallint,
ss_snm_acc smallint,
ss_snm_prot smallint,
ss_class_prot smallint,
ss_prop_prot smallint,
ss_hostile_prot smallint,
ss_infr smallint,
mi_comp smallint,
mi_tech smallint,
mi_bb smallint,
mi_mission smallint,
mi_asset smallint,
mi_natl smallint,
mi_infr smallint
);


--
-- TOC entry 8 (OID 37311)
-- Name: camp_score; Type: ACL; Schema: public; Owner: postgres
--

REVOKE ALL ON TABLE camp_score FROM PUBLIC;
GRANT INSERT, SELECT, UPDATE ON TABLE camp_score TO mn;
GRANT ALL ON TABLE camp_score TO mndb;
GRANT SELECT ON TABLE camp_score TO mgr;
GRANT INSERT, SELECT, UPDATE ON TABLE camp_score TO mnview;



---------------------------(end of broadcast)---------------------------
TIP 4: Don't 'kill -9' the postmaster

"PostgreSQL Bugs List" <pgsql-bugs (AT) postgresql (DOT) org> writes:
Doesn't happen for anybody else...

I'm wondering about a buggy compiler. The code involved is much too
simple to be broken: it's repeated calls to this:

static void
AddAcl(char *aclbuf, const char *keyword)
{
if (*aclbuf)
strcat(aclbuf, ",");
strcat(aclbuf, keyword);
}

If *aclbuf contains zero on entry, then why isn't the strcat overwriting
instead of appending? Looks like a compiler bug to me. What is "-fast"
and does the problem go away if you don't use it?

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
joining column's datatypes do not match