[GENERAL] Is there an md5sum for tables?

Hi all, I need to know if multiple tables have (may have most probably) identical content. Since I want a fast solution (which means not comparing tables row by row), I thought it would be a good idea to have an sql function operating on a table or view similar to md5sum on a file and only compa

[GENERAL] util/int8.h: missing int8_text() function

Hi all, I migrate from 8.1 to 8.3. In 8.2.7 and previous there are functions int4_text() and int8_text(), in 8.3.0 they are missing (not documented that they are removed in the release notes). What I'm supposed to use instead of int8_text() now? (I use my own datatype which is similar to varchar/

[GENERAL] good experience with performance in 8.2 for multi column indexes

Hi lists, I want to let you take part in my experience of performance boost for delete operations where more than one column is part of a primary key. For my setup, in 8.1 a delete query which deletes 20 entries depending on rows in another table runs about 7h, in 8.2 (and later) it runs 9s

Re: [GENERAL] util/int8.h: missing int8_text() function

Alvaro Herrera wrote: Michael Enke wrote: Hi all, I migrate from 8.1 to 8.3. In 8.2.7 and previous there are functions int4_text() and int8_text(), in 8.3.0 they are missing (not documented that they are removed in the release notes). What I'm supposed to use instead of int8_text() no

Re: [GENERAL] good experience with performance in 8.2 for multi column indexes

No, both negative. Michael Tom Lane wrote: Michael Enke <[EMAIL PROTECTED]> writes: For my setup, in 8.1 a delete query which deletes 20 entries depending on rows in another table runs about 7h, in 8.2 (and later) it runs 9s! If those rowcount estimates are even close to ac

[GENERAL] CREATE CHARSET would be nice feature

Hi all, I had to get data out of postgresql in CP852 encoding (internally stored in UNICODE). So I wrote my own encoding function and used CREATE DEFAULT ENCODING. The problem with this is: I can only use character set names which are known internally. So I "misused" the LATIN2: update pg_conver

[GENERAL] pg_dumpall: pg_conversion table not saved

Hi list, I updated pg_converion to set the condefault=false where I need it. pg_dumpall does NOT write that table out and therefore it is not possible to restore, this update get lost. Anybody know if this is an error or has an explanation why it has to work this way? Background: I have created

Re: [GENERAL] pg_dumpall: pg_conversion table not saved

Tom Lane schrieb: > Michael Enke <[EMAIL PROTECTED]> writes: >> I updated pg_converion to set the condefault=false where I need it. > > Why are you needing to replace an existing default conversion? > Do we have a bug in it? No, not a bug. But I need CP852 encoding conver

[GENERAL] is there a 'replace' for tables/views?

Hello group, is there a functionality like 'create or REPLACE table/view' to circumvent drop with 'cascade' and recreation of depending objects? I have a table schema1.x (col1, col2) and a view schema1.v -> schema1.x(col1) Another table is schema2.x(col1, col2) with same structure as x in schema

[GENERAL] PG_MODULE_MAGIC lost if strip the object file

Hi list, I create a dynamic library with the PG_MODULE_MAGIC under Linux. No problem in normal situation, but if I strip the .o file, that information seems to get lost and I get: ERROR: incompatible library "...so": missing magic block TIP: Extension libraries are required to use the PG_MODULE_

Re: [GENERAL] PG_MODULE_MAGIC lost if strip the object file

Alvaro Herrera wrote: Michael Enke wrote: Hi list, I create a dynamic library with the PG_MODULE_MAGIC under Linux. No problem in normal situation, but if I strip the .o file, that information seems to get lost and I get: ERROR: incompatible library "...so": missing magic