Date: Wed, 7 Sep 2005 17:04:30 +0000
From: nathan wagner <nw@hydaspes.if.org>
To: pgsql-hackers@postgresql.org
Subject: Re: uuid type for postgres
Message-ID: <20050907170430.GA6281@granicus.if.org>
References: <20050906135057.GA24136@granicus.if.org>
	<7CDBCE29-55BE-4EE3-8868-F50DB1E914F6@refractions.net>
	<20050907051508.GA3572@granicus.if.org>
	<200509070945.17916.josh@agliodbs.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <200509070945.17916.josh@agliodbs.com>
User-Agent: Mutt/1.4.2.1i

On Wed, Sep 07, 2005 at 09:45:17AM -0700, josh@agliodbs.com wrote:

> I think the issue is portability.  Remember that this type needs to work on 
> Windows as well as all POSIX platforms and AIX.

I had forgotten about windows.  I'll see to what extent the library i'm
using is portable to windows.

> > It would appear that linking against libraries is just fine, I don't
> > see that uuid is any different.
> 
> Is it portable to all platforms we support?

I don't know yet.  It could be made to be so though.  I don't have any way
to compile on windows, so someone would have to be willing to try that.
Same with AIX.  If someone with an AIX box could download and test
it out I would appreciate a compatibility report.

> How does it generate its machine identifier?

No idea.  Does it matter?  Not having to fret this kind of detail is
the advantage of using someone else's library.

> One of the differences between an add-in and core code is support for all 
> PostgreSQL platforms.

So would including my type in contrib be possible then?  To me, that
is perhaps more important than making it a core type, though i think
there should be a core UUID type as well.

I'm testing out elog().  I'll post a new version later today.  There really
won't be any new functionality, it's just code clean up.

-- 
Nathan Wagner