Tom Lane wrote:
> This isn't a bug. The DELETE causes the a=0 row to disappear from the
When I got choice between rules and triggers, I looked carefully is
documantation, section "33.6. Rules versus Triggers". As I can see,
rules prefered, there are many examples, why rules better then triggers.
Only described advantage of triggers is using triggers as constraints.
But later I got bugs and submit #1447, #1610, #1676. All resolved as
"not a bug". IMHO this is due to lack of documentation, don't described
explicitly what disadvatages or unexpected "features" rules have. Well
#1676 looked strange, but can de derived from "33.3. Rules on INSERT,
UPDATE, and DELETE". As I can undestand, if rule fire several commands,
command to change table of rule must be last. This is not explicitly
documented. ALSO rules didn't designed so. Why?
As I begin undestand now, bug #1447 is consequence from "feature" #1676,
isn't it? And can be avoided by changing ALSO rule to INSTEAD rule with
command to modify table, implicitly existed in ALSO rule, in the last.
This workaround must be documented. Or may be better change ALSO rule
#1610 disconform documentation. Documentation or rules must be fixed.
Fix rules prefered.
As I undestand from bug resolution, this "features" don't planed to be
fixed. I add comments to documentation about this "features", but this
is all, what I can. As I can undestand now, my comment in documentation
for "feature" #1447 in section "33.6. Rules versus Triggers" is
incorrect, but I can't remove it.
In response to
pgsql-bugs by date
|Next:||From: Tom Lane||Date: 2005-05-20 15:40:05|
|Subject: Re: CHECK constraint (true) causes dumps with parse errors |
|Previous:||From: Neil Conway||Date: 2005-05-20 08:22:21|
|Subject: Re: CHECK constraint (true) causes dumps with parse errors|