 | |
"name order was judged to be more convenient" (multiple triggers)
comp.databases.postgresql comp.databases.postgresql
 |
| | Thread Tools | Display Modes |
#1
 
| |||
| |||
|
"name order was judged to be more convenient" (multiple triggers) -
11-11-2009
, 05:58 AM
|
SQL specifies that multiple triggers should be fired in time-of-creation order. PostgreSQL uses name order, which was judged to be more convenient. |
One of the basic rules in programming/coding is that name of class/
variable/function/etc. should describe its purpose/function/use. Being
forced to adjust name of object to manage its position within some
kind of stack is against this concept.
Obviously I can't tell what _exactly_ was meant by "more convenient"
as it depends on context, but from my (user) point of view the time-of-
creation order sounds to be more clear/proper approach, re-creating
triggers in desired order is a matter of seconds.
Best regards
J.Záruba
#2
| |||
| |||
|
|
SQL specifies that multiple triggers should be fired in time-of-creation order. PostgreSQL uses name order, which was judged to be more convenient. I have to say that this seems wrong to me. One of the basic rules in programming/coding is that name of class/ variable/function/etc. should describe its purpose/function/use. Being forced to adjust name of object to manage its position within some kind of stack is against this concept. Obviously I can't tell what _exactly_ was meant by "more convenient" as it depends on context, but from my (user) point of view the time-of- creation order sounds to be more clear/proper approach, re-creating triggers in desired order is a matter of seconds. |
Currently the trigger creation time is not stored in the catalogs.
Also, I think it is pretty inconvenient if the trigger firing order
changes if I drop and recreate a trigger...
Alphabetical ordering is a simple rule; if you care about
trigger order, you just follow the rule.
Just for the record, this is how Oracle solves it:
Oracle fires multiple triggers in an unspecified,
random order, if more than one trigger of the same type
exists for a given statement; that is, triggers of the
same type for the same statement are not guaranteed
to fire in any specific order.
(from http://download.oracle.com/docs/cd/B...tm#sthref3278).
Yours,
Laurenz Albe
#3
| |||||
| |||||
|
|
Jaroslav Záruba wrote: SQL specifies that multiple triggers should be fired in time-of-creation order. PostgreSQL uses name order, which was judged to be more convenient. I have to say that this seems wrong to me. One of the basic rules in programming/coding is that name of class/ variable/function/etc. should describe its purpose/function/use. Being forced to adjust name of object to manage its position within some kind of stack is against this concept. Obviously I can't tell what _exactly_ was meant by "more convenient" as it depends on context, but from my (user) point of view the time-of- creation order sounds to be more clear/proper approach, re-creating triggers in desired order is a matter of seconds. One convenience aspect clearly is the convenience fo the developers: Currently the trigger creation time is not stored in the catalogs. |
Quote:
|
Also, I think it is pretty inconvenient if the trigger firing order changes if I drop and recreate a trigger... |
is number of your triggers. Maybe I'm just obsessed with proper
naming.
)But given the creation time does not get stored it is kinda
irrelevant.
Quote:
|
Alphabetical ordering is a simple rule; if you care about trigger order, you just follow the rule. |
Quote:
|
Just for the record, this is how Oracle solves it: * Oracle fires multiple triggers in an unspecified, * random order, if more than one trigger of the same type * exists for a given statement; that is, triggers of the * same type for the same statement are not guaranteed * to fire in any specific order. (fromhttp://download.oracle.com/docs/cd/B19306_01/server.102/b14220/trigge...). |
I love Postgres.
Quote:
|
Yours, Laurenz Albe |
JZ
#4
| |||
| |||
|
|
SQL specifies that multiple triggers should be fired in time-of-creation order. PostgreSQL uses name order, which was judged to be more convenient. I have to say that this seems wrong to me. One convenience aspect clearly is the convenience fo the developers: Currently the trigger creation time is not stored in the catalogs. Aha, that would be it. |
It just seems that nobody saw any benefits in that.
I searched the archives and didn't find any discussion.
I guess nobody had a problem with it...
I found the commit message where this was added:
http://archives.postgresql.org/pgsql...4/msg00149.php
It looks like this was a side effect.
Yours,
Laurenz Albe
|
« Previous Thread
|
Next Thread »
| Thread Tools | |
| Display Modes | |
Linear Mode
| |
Powered by vBulletin Version 3.5.3
Copyright ©2000 - 2012, Jelsoft Enterprises Ltd.
Copyright ©2000 - 2012, Jelsoft Enterprises Ltd.