From: "Oliver Jowett"
If the server is shut down mid-query, doesn't the backend complete the
current query cycle before closing the connection?
i.e. we'd see ErrorResponse, ReadyForQuery, and return control to the
app before seeing EOF anyway?
The protocol spec is a bit vague there.
From an o
be "STABLE" instead of "VOLATILE", then
the problem goes away.
Regards
Donald Fraser
;
I also noted that if I replace the function call in the view with equivalent
SQL:
(SELECT (CASE WHEN length(c.s_umbname) > 0 THEN c.s_umbname || ' - '::citext ||
c.s_name ELSE c.s_name END) AS s_cmpyname FROM tbl_cmpy c WHERE c.id =
n.id_contrib LIMIT 1) AS s_provider
the problem also goes away.
Regards
Donald Fraser
spx
and a further announcement will be made once the annual financial report
is available to be downloaded. "
Regards
Donald Fraser
nThis is a message on a new line
4)
select test_func_exception4();
ERROR: This is an error message.
This is a message on a new line
You will note that even using the new E'' string format syntax for the RAISE
EXCEPTION appears to be broken (test_func_exception3()).
I can't find anything in the documentation that suggests this should be the
observed behaviour.
Regards
Donald Fraser
len; name_len--
63(1 row)
I don't know whether you consider this a bug, certainly most
of the other data types appear to have range checking as part of the
input.
Regards
Donald Fraser
ectory is set via:
%{?build7x:%define kerbdir
/usr/kerberos}
%{?build89:%define kerbdir
/usr/kerberos}
where as the 7.4.6 version has the kerberos
directory is set via:
%{?build7x:%define kerbdir "/usr"}
%{?build89:%define kerbdir "/usr"}
I think you'll agree that this directory change is not the default for
Redhat 7, 8 or 9
Regards
Donald Fraser
ion
case 'N': // Error Notification
int l_nlen = pgStream.ReceiveIntegerR(4);
PSQLException notify =
PSQLException.parseServerError(connection.getEncoding().decode(pgStream.Receive
(l_nlen-4)));
statement.addWarning(notify.getMessage());
break;
Re
ver come this problem I had to change the dynamic query as such:
squery := ''SELECT text('' || scolname || '') AS s_data FROM '' || stblname ||
'' WHERE id = '' || idrow || '' LIMIT 1'';
FOR data_rec IN
this problem on version 7.3.4 which is what I have upgraded
from.
Regards
Donald
Fraser
blem on version 7.3.4 which is what I have upgraded
from.
Regards
Donald Fraser
: permission denied for relation vu_tbl_useracc
The error goes away if I grant both INSERT and
UPDATE permissions to the above group.
I didn't have this problem on version 7.3.4 which is what I have upgraded
from.
Regards
Donald Fraser
_key ON
tbl_test USING btree (s_mnem) WHERE ((n_status
& -10) <> 0);
Note: the binary inverse of 9 is -10.
Explain shows that the planner now chooses to
use the partial index?
Regards
Donald Fraser.
OLUMN id_security TO id_seclass;
The Rule stated above never changed.
I noted that the underlying SELECT rule (rule name _RETURN) for the view
changed by replacing the column named "id_security" to "id_seclass AS
id_security".
Regards
Donald Fraser
Ps. The way I checked that the Rule
g the correct syslog facility setting as
specified in the configuration file.
Regards
Donald Fraser.
ERRABLE INITIALLY DEFERRED then we can achieve the scenario
where Transaction B can complete before Transaction A.
It still doesn't explain why a RowExclusive lock is
required on the foreign key table, when we want be updating
it?
Apologies in advance to those experienced users if
I have overlooked something fundamental and obvious.
Thanks
Donald Fraser.
> "Donald Fraser" <[EMAIL PROTECTED]> writes:
> > What would be nice is when you don't specify a default on the view, the
default
> > from the table is used.
>
> What table? You seem to have an extremely narrow-minded concept of what
> a view
> "Donald Fraser" <[EMAIL PROTECTED]> writes:
> > I have a major problem with DEAFAULT values:
>
> Consider attaching a default to the view column. As-is, its default
> is NULL (and I can't see any good argument for the factory default
> being anyth
straight tables, as the client software can easily
avoid INSERTS with NULLs, but this situation is totally unusable for
views because you don't know whether the user was supplying a NULL or
whether it simply wasn't specified. Even if did want to avoid this NULL
situation the number
- Original Message -
From: "Tom Lane" <[EMAIL PROTECTED]>
To: "Donald Fraser" <[EMAIL PROTECTED]>
Cc: "[BUGS]" <[EMAIL PROTECTED]>
Sent: Monday, February 24, 2003 8:29
PM
Subject: Re: [BUGS] Bugs with rules on
views/tables: permiss
- Original Message
- From: "Tom Lane" <[EMAIL PROTECTED]>To: "Donald Fraser"
<[EMAIL PROTECTED]>Cc: "[ADMIN]"
<[EMAIL PROTECTED]>Sent: Monday, February 24, 2003 2:42
PMSubject: Re: [ADMIN] Bugs with rules on views/tables: permis
-
Original Message - From: "Tim Burgess" <[EMAIL PROTECTED]>To: <[EMAIL PROTECTED]>Sent: Monday, February 10, 2003 6:51
PMSubject: [BUGS] 'update' as action of 'insert' rule: permission
denied> Gday All,> > I'm surprised I can't find any
descriptions of this in the archive > already,
- Original Message -
From:
Donald Fraser
To: [EMAIL PROTECTED]
Sent: Thursday, February 06, 2003 11:27
AM
Subject: [BUGS] Trigger function not
executing
OS: Redhat, kernel 2.4
PostgreSQL 7.3.1 on i686-pc-linux-gnu, compiled
by GCC 2.96
I have
the function without setting the flag I
have mentioned.
Have I misunderstood trigger functions or is this a
legitimate bug?
Regards
Donald Fraser
24 matches
Mail list logo
