Hi,
I hit a UPDATE/LOCK issue in my application and the result has surprised me
somewhat.
And for the repro, it boils down into this:
---
CREATE TABLE x (a int, b bool);
INSERT INTO x VALUES (1, TRUE);
COMMIT;
_THREAD 1_:
BEGIN;
UPDATE x SET b=FALSE;
INSERT INTO x VALUES (2, TRUE);
Hmm...
I tend to _mostly_ run workstations rather than servers, & pick my distro to
suit my application needs.
My workplace is a SLES site, & I use Open Suse. Given most of my Postgres
databases are in fact PostGIS databases, and need to work with a variety of
other spatial data & GIS related
On 04/03/12 09:49, John R Pierce wrote:
On 03/03/12 2:55 AM, Gavin Flower wrote:
My knowledge of Debian is via friend's (an extremely competent and
experienced Unix guy who got me into Linux & who still runs Debian)
comments and what I've noticed on the web. For a Desktop development
machin
Greetings to all.
I want to know if with postgis 2.0 and postgresql is possible to generate a
topology from the geometries contained inside a shape file. If this can be
accomplished, where can I find the procedure i must follow?
Thanks in advance.
--
Sent via pgsql-general mailing list (pgs
On Mar 5, 2012, at 0:08, Chris Angelico wrote:
> On Mon, Mar 5, 2012 at 3:15 PM, Tom Molesworth wrote:
>> Can you use to_number() here? It sounds like something along the lines of
>> cast(to_number('0' || field::varchar, '9.') as int) might give the
>> behaviour you're after, and a quick
On Mon, Mar 5, 2012 at 3:15 PM, Tom Molesworth wrote:
> Can you use to_number() here? It sounds like something along the lines of
> cast(to_number('0' || field::varchar, '9.') as int) might give the
> behaviour you're after, and a quick test seems to indicate that it's about
> 4x faster th
On 05/03/12 04:06, Chris Angelico wrote:
On Mon, Mar 5, 2012 at 2:50 PM, David Johnston wrote:
Any efficient, non-RegEx, alternative would require more context to evaluate
than you provide. Mainly, would it be faster to have a separate field to store
the parsed (at input) number and then que
On Mon, Mar 5, 2012 at 2:50 PM, David Johnston wrote:
> Any efficient, non-RegEx, alternative would require more context to evaluate
> than you provide. Mainly, would it be faster to have a separate field to
> store the parsed (at input) number and then query that field directly (even
> if it
On Mar 4, 2012, at 22:31, Chris Angelico wrote:
> (Hoping you meant for that reply to be on-list as I'm here responding
> on-list.)
>
> On Mon, Mar 5, 2012 at 2:16 PM, A.M. wrote:
>>
>> On Mar 4, 2012, at 9:13 PM, Chris Angelico wrote:
>>
>>> One of our tables has a few columns that may be i
(Hoping you meant for that reply to be on-list as I'm here responding on-list.)
On Mon, Mar 5, 2012 at 2:16 PM, A.M. wrote:
>
> On Mar 4, 2012, at 9:13 PM, Chris Angelico wrote:
>
>> One of our tables has a few columns that may be interpreted as strings
>> or may be numbers (data type is varchar,
One of our tables has a few columns that may be interpreted as strings
or may be numbers (data type is varchar, numbers are stored as
decimal). Generally, operations are performed on the string, but
sometimes we need to parse out a number - without it failing on error.
I wrote the following functio
On 04/03/12, Rory Campbell-Lange (r...@campbell-lange.net) wrote:
> On 04/03/12, Jan Meyland Andersen (j...@agile.dk) wrote:
> > My main problem is that I do not know how many columns or the data
> > type of the columns before runtime.
> > It this possible at all?
> There is a section on this in
On 04/03/12, Jan Meyland Andersen (j...@agile.dk) wrote:
> How do I return an unknown resultset from a function
>
> My main problem is that I do not know how many columns or the data
> type of the columns before runtime.
> It this possible at all?
>
> I also tried to return the data as a text arr
c k writes:
> what will be the reason for this?
The short answer is that || uses cast-to-text semantics, whereas concat
uses output-function semantics, and char(n) is one of the weird types
for which those are different. Don't blame us, blame the SQL committee.
Or rather than blaming anybody, st
On Mar 4, 2012, at 14:52, Jan Meyland Andersen wrote:
> How do I return an unknown resultset from a function
>
> My main problem is that I do not know how many columns or the data type of
> the columns before runtime.
> It this possible at all?
>
> I also tried to return the data as a text a
Hello
2012/3/4 Jan Meyland Andersen :
> How do I return an unknown resultset from a function
>
> My main problem is that I do not know how many columns or the data type of
> the columns before runtime.
> It this possible at all?
>
no, or it is not possible simply. PostgreSQL is strongly typed.
I
How do I return an unknown resultset from a function
My main problem is that I do not know how many columns or the data type
of the columns before runtime.
It this possible at all?
I also tried to return the data as a text array but I also have trouble
with that.
Regards
Jan
--
Sent via p
Both varchar and text keep the trailing whitespace since it must be
significant. A char removes all trailing whitespace (with the || operator)
since it cannot distinguish between padding supplied whitespace (insignificant)
and user-supplied whitespace (significant).
On Mar 4, 2012, at 13:26, c
I have found an another difference.
select 'AA '::char(5) || 'AA'::char(5);
result: ""
while
select 'AA '::varchar(5) || 'AA'::varchar(5);
gives
result: "AA AA"
select 'AA '::text || 'AA'::text;
gives
result: "AA AA"
what will be the reason for this?
Regards,
C P Kulkarni
On
Greetings to all.
I want to know if with postgis 2.0 and postgresql is possible to generate a
topology from the geometries contained inside a shape file. If this can be
accomplished, where can I find the procedure i must follow?
Thanks in advance.
signature.asc
Description: This is a digitally
On Sat, 2012-03-03 at 14:15 -0700, David Boreham wrote:
>
> We use CentOS 5 and 6 and install PG from the yum repository detailed
> on the postgresql.org web site.
Those RPMs will probably be a part of CentOS Testing repository soon. I
and Karanbir had a chat about it at FOSDEM this year.
--
D
21 matches
Mail list logo
