pg.ddx.io pgsql-interfaces@postgresql.org mailing list archive
help / color / mirror / Atom feedsequences
7+ messages / 6 participants
[nested] [flat]
* sequences
@ 1999-04-29 18:05 Patrick Welche <prlw1@newn.cam.ac.uk>
0 siblings, 0 replies; 7+ messages in thread
From: Patrick Welche @ 1999-04-29 18:05 UTC (permalink / raw)
To: pgsql-interfaces
After the interesting discussion on what is the safest way to have a unique
"oid"-like value per row, I thought I would have a go, but from libpq++ I
don't seem to be able to read the new sequence value:
Table = person
+----------------------------------+----------------------------------+-------+
| Field | Type | Length|
+----------------------------------+----------------------------------+-------+
| id | int4 not null default nextval('p | 4 |
| surname | text not null | var |
| firstname | text not null | var |
| email | text | var |
| phone | text | var |
| rfbdate | date | 4 |
+----------------------------------+----------------------------------+-------+
Index: person_id_key
The output from my test program is:
INSERT INTO person (surname,firstname,email,phone) VALUES ('Welche','Patrick','
prlw1',null)
SELECT currval('person_id_seq')
Missing person id
and the relevant bit of code is:
query<<"INSERT INTO person (surname,firstname,email,phone) VALUES ('"
<<a._surname<<"','"<<a._firstname<<"',"
<<ISNULL(a._email)<<','
<<ISNULL(a._phone)<<')'<<ends;
s.send_query(query);
query<<"SELECT currval('person_id_seq')"<<ends;
s.send_query(query);
if(s.Tuples()!=1)throw empty_result("Missing person id");
a.id(toid(s.GetValue(0,"currval")));
send_query prints the query and does a Exec(query). I suppose I don't
understand the difference between doing things interactively in psql (the
above INSERT and SELECT work then) and calling things from a program. Any
tips?
Cheers,
Patrick
^ permalink raw reply [nested|flat] 7+ messages in thread
* RE: [INTERFACES] sequences
@ 1999-04-29 19:41 Jackson, DeJuan <djackson@cpsgroup.com>
0 siblings, 1 reply; 7+ messages in thread
From: Jackson, DeJuan @ 1999-04-29 19:41 UTC (permalink / raw)
To: prlw1@cam.ac.uk; pgsql-interfaces
You might try placing the INSERT and SELECT CURRVAL into a transaction
block.
-DEJ
> -----Original Message-----
> From: Patrick Welche [SMTP:prlw1@newn.cam.ac.uk]
> Sent: Thursday, April 29, 1999 1:05 PM
> To: pgsql-interfaces@postgreSQL.org
> Subject: [INTERFACES] sequences
>
> After the interesting discussion on what is the safest way to have a
> unique
> "oid"-like value per row, I thought I would have a go, but from libpq++ I
> don't seem to be able to read the new sequence value:
>
> Table = person
> +----------------------------------+----------------------------------+---
> ----+
> | Field | Type |
> Length|
> +----------------------------------+----------------------------------+---
> ----+
> | id | int4 not null default nextval('p |
> 4 |
> | surname | text not null |
> var |
> | firstname | text not null |
> var |
> | email | text |
> var |
> | phone | text |
> var |
> | rfbdate | date |
> 4 |
> +----------------------------------+----------------------------------+---
> ----+
> Index: person_id_key
>
> The output from my test program is:
>
> INSERT INTO person (surname,firstname,email,phone) VALUES
> ('Welche','Patrick','
> prlw1',null)
> SELECT currval('person_id_seq')
>
> Missing person id
>
> and the relevant bit of code is:
>
> query<<"INSERT INTO person (surname,firstname,email,phone) VALUES
> ('"
> <<a._surname<<"','"<<a._firstname<<"',"
> <<ISNULL(a._email)<<','
> <<ISNULL(a._phone)<<')'<<ends;
> s.send_query(query);
> query<<"SELECT currval('person_id_seq')"<<ends;
> s.send_query(query);
> if(s.Tuples()!=1)throw empty_result("Missing person id");
> a.id(toid(s.GetValue(0,"currval")));
>
> send_query prints the query and does a Exec(query). I suppose I don't
> understand the difference between doing things interactively in psql (the
> above INSERT and SELECT work then) and calling things from a program. Any
> tips?
>
> Cheers,
>
> Patrick
^ permalink raw reply [nested|flat] 7+ messages in thread
* RE: [INTERFACES] sequences
@ 1999-05-03 11:57 Herouth Maoz <herouth@oumail.openu.ac.il>
parent: Jackson, DeJuan <djackson@cpsgroup.com>
0 siblings, 1 reply; 7+ messages in thread
From: Herouth Maoz @ 1999-05-03 11:57 UTC (permalink / raw)
To: Jackson, DeJuan <djackson@cpsgroup.com>; prlw1@cam.ac.uk; pgsql-interfaces
At 22:41 +0300 on 29/04/1999, Jackson, DeJuan wrote:
> You might try placing the INSERT and SELECT CURRVAL into a transaction
> block.
That shouldn't make a difference. Currval's scope is the entire session,
from the last call (or implicit call) to nextval to the next call.
Now, let's see if I understand right: The query works well in psql, but not
in a program? So let me ask you this - do you run the program, perhaps, as
a different user, such as the httpd or nobody user? If so, you need to
GRANT SELECT on the sequence to be able to use the currval function.
Herouth
--
Herouth Maoz, Internet developer.
Open University of Israel - Telem project
http://telem.openu.ac.il/~herutma
^ permalink raw reply [nested|flat] 7+ messages in thread
* Re: [INTERFACES] sequences
@ 1999-05-04 16:11 Patrick Welche <prlw1@newn.cam.ac.uk>
parent: Herouth Maoz <herouth@oumail.openu.ac.il>
0 siblings, 0 replies; 7+ messages in thread
From: Patrick Welche @ 1999-05-04 16:11 UTC (permalink / raw)
To: herouth@oumail.openu.ac.il; +Cc: pgsql-interfaces
Herouth Maoz wrote:
>
> At 22:41 +0300 on 29/04/1999, Jackson, DeJuan wrote:
>
>
> > You might try placing the INSERT and SELECT CURRVAL into a transaction
> > block.
>
> That shouldn't make a difference. Currval's scope is the entire session,
> from the last call (or implicit call) to nextval to the next call.
>
> Now, let's see if I understand right: The query works well in psql, but not
> in a program? So let me ask you this - do you run the program, perhaps, as
> a different user, such as the httpd or nobody user? If so, you need to
> GRANT SELECT on the sequence to be able to use the currval function.
Thank you!! That was it. (I had granted access to tables, not to sequences)
Cheers,
Patrick
^ permalink raw reply [nested|flat] 7+ messages in thread
* sequences
@ 1999-11-02 11:01 StOo <stuart@comodo.net>
0 siblings, 1 reply; 7+ messages in thread
From: StOo @ 1999-11-02 11:01 UTC (permalink / raw)
To: pgsql-interfaces
Hi,
The size of a sequence is 32bit , correct? (im pretty damn sure it is). Is
there any way I can tune this to 64bit, There is an int8 option in postgres as
I use that already.Threre is a skanky way of creating an incremenetd counter
whilst using an int8 but I would rather up the sequence in the code. Anyone got
any ideas where to look?
I am aware that the int8 is signed but that doest matter currently.
+-=-=-=-=-=-=-=StOo=-=-=-=-=-=-=-+
Stuart Hodgkinson Stuart@comodo.net
^ permalink raw reply [nested|flat] 7+ messages in thread
* sequences
@ 1999-11-03 15:32 stuart <stuart@comodo.net>
0 siblings, 0 replies; 7+ messages in thread
From: stuart @ 1999-11-03 15:32 UTC (permalink / raw)
To: pgsql-interfaces
Hi,
The size of a sequence is 32bit , correct? (im pretty damn sure it is). Is
there any way I can tune this to 64bit, There is an int8 option in postgres
as
I use that already.Threre is a skanky way of creating an incremenetd
counter
whilst using an int8 but I would rather up the sequence in the code. Anyone
got
any ideas where to look?
I am aware that the int8 is signed but that doest matter currently
Stuart Hodgkinson Stuart@comodo.net
^ permalink raw reply [nested|flat] 7+ messages in thread
* Re: [INTERFACES] sequences
@ 1999-11-04 15:53 Thomas Lockhart <lockhart@alumni.caltech.edu>
parent: StOo <stuart@comodo.net>
0 siblings, 0 replies; 7+ messages in thread
From: Thomas Lockhart @ 1999-11-04 15:53 UTC (permalink / raw)
To: stuart@comodo.net; +Cc: pgsql-interfaces
> The size of a sequence is 32bit , correct? (im pretty damn sure it is). Is
> there any way I can tune this to 64bit, There is an int8 option in postgres as
> I use that already.Threre is a skanky way of creating an incremenetd counter
> whilst using an int8 but I would rather up the sequence in the code. Anyone got
> any ideas where to look?
Look in src/backend/parser/gram.y and analyze.c for where sequences
(the serial type) is handled. The code actually just embeds the
"skanky" technique to hide it from the user. A very first cut might be
to make this configurable by defining, say, SERIAL_IS_INT64. Patches
accepted ;)
- Thomas
--
Thomas Lockhart lockhart@alumni.caltech.edu
South Pasadena, California
^ permalink raw reply [nested|flat] 7+ messages in thread
end of thread, other threads:[~1999-11-04 15:53 UTC | newest]
Thread overview: 7+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
1999-04-29 18:05 sequences Patrick Welche <prlw1@newn.cam.ac.uk>
1999-04-29 19:41 RE: [INTERFACES] sequences Jackson, DeJuan <djackson@cpsgroup.com>
1999-05-03 11:57 ` Herouth Maoz <herouth@oumail.openu.ac.il>
1999-05-04 16:11 ` Patrick Welche <prlw1@newn.cam.ac.uk>
1999-11-02 11:01 sequences StOo <stuart@comodo.net>
1999-11-04 15:53 ` Thomas Lockhart <lockhart@alumni.caltech.edu>
1999-11-03 15:32 sequences stuart <stuart@comodo.net>
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox