Re: Views, views, views: Summary of Arguments - Mailing list pgsql-hackers

From Robert Treat
Subject Re: Views, views, views: Summary of Arguments
Date
Msg-id 200505122206.39685.xzilla@users.sourceforge.net
Whole thread Raw
In response to Re: Views, views, views: Summary of Arguments  (David Fetter <david@fetter.org>)
List pgsql-hackers
On Thursday 12 May 2005 19:59, David Fetter wrote:
> On Thu, May 12, 2005 at 05:52:43PM -0400, Robert Treat wrote:
> > On Thursday 12 May 2005 01:32, Christopher Kings-Lynne wrote:
> > > > FWIW, I don't see the issue as "internal vs external" at all.
> > > > What's bothering me is whether these views can be considered
> > > > sufficiently more stable and better designed than the physical
> > > > system catalogs to justify recommending that application
> > > > designers should rely on the views instead of the catalogs.
> > > > That point doesn't seem to me to have been proven.  The recent
> > > > arguments in favor seem to boil down to "novices will find these
> > > > easier to use", which is very possibly true, but novices don't
> > > > have the same needs as programs.
> > >
> > > As lead phpPgAdmin developer, I'm officially in favour of them.
> > > The main reason being all the extra fruit they have that shows
> > > database size, etc.
> >
> > As non-lead phpPgAdmin developer, I'd be against using them in
> > phppgadmin.  (note this doesnt mean I am against them in pgsql
> > itself)
>
> On what grounds?
>

See upthread response from the various pgadmin guys as to why they wouldn't 
use them. 

> > > That means we can display this meta information in phpPgAdmin and
> > > not worry about having to re-implement it all.
> >
> > You have to query something...  doesn't bother me to go to the
> > source.
>
> A PostgreSQL developer has shown in this very thread that it is
> extremely easy to screw up a query against those catalogs.  Maybe
> you're better than he is, but that's not a reason to keep something
> simpler out.
>

Out of phppgadmin or out of pgsql itself?  The internals of phppgadmin dont 
have to be simple (sure its a bonus when they are, but its not the primary 
goal).  Of course thats completly orthagonal to whether these views should be 
in pgsql itself. Unless you think they should be in there for admin tools. 
Personally I think at least a couple of these should be in there for the ease 
of use factor (relacl is just evil), but the other reasons of unchanging 
interface and defining all objects seem wrong headed to me and make me 
skeptical of the whole thing. 

-- 
Robert Treat
Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL


pgsql-hackers by date:

Previous
From: Andrew Dunstan
Date:
Subject: Re: Views, views, views: Summary of Arguments
Next
From: "Greg Sabino Mullane"
Date:
Subject: libpq lo_open errors when first action in connection