On 5/21/15 7:12 AM, Robert Haas wrote:
> On Wed, May 20, 2015 at 8:46 PM, Andres Freund <andres@anarazel.de> wrote:
>> I've a hard time believing it's actually a good idea to change this. It
>> pretty much seems to only be useful if you're doing unqualified SELECT
>> pg_cancel_backend(pid) FROM pg_stat_activity; type queries. I don't see
>> that as something we need to address.
>
> +1. I'm not saying this isn't annoying - I've been annoyed by it
> myself - but IMHO it's really not worth having two functions that do
> 99% the same thing. Then, instead of having to remember to exclude
> your own backend using the same SQL syntax you use for everything
> else, you have to remember which of two similarly-named functions to
> call if you don't want to kill your own backend. That might be better
> for some people, but it won't be better for everyone.
OTOH, if we allowed RAISE FATAL in plpgsql there'd be no need for
self-termination via pg_terminate_backend(). I also don't see a problem
with allowing self-termination, with the default being to disallow it.
I suspect the number of people writing code that need self-termination
is very, very small, whereas probably every DBA out there has been
bitten by this. This is the type of thing that gives Postgres a
reputation for being 'hard to use'. I would think the benefit of the
larger group would outweigh the pain the smaller group would feel
changing their code, but of course that's just my opinion and I don't
see any easy way to quantify that.
You and Andreas think it's fine as-is.
Tom and Jon Nelson maybe don't like it as-is, but won't break backwards
compatibility.
David Steele and I seem fine with breaking compat., not sure about Fabrizio.
Given the opposition unless some others speak up I'll just drop it.
--
Jim Nasby, Data Architect, Blue Treble Consulting
Data in Trouble? Get it in Treble! http://BlueTreble.com