On Thu, Nov 24, 2005 at 02:46:50PM -0800, Darren Duncan wrote:
> At 9:39 PM +0000 11/24/05, Tim Bunce wrote:
> >On Wed, Nov 02, 2005 at 06:28:00PM -0800, Darren Duncan wrote:
> >> Today when I was on #perl6 at irc.freenode.net and, out of the blue
> >> just following my connect, autrijus asked me if I would port
> >> DBI::PurePerl and DBD::MySQLPP to Perl 6 such that their presence can
> >> help with Pugs new object handling abilities.
> >
> >I think that's a bad idea, personally.
> >
> >I see little good and much harm that could come from doing this.
> >
> >I want DBI v1 to be a poster child for the ability of Perl 6 to run
> >Perl 5 extensions - one way or another (either via ponie or an embedded
> >libperl5.so).
> >
> >The goal of helping "Pugs new object handling abilities" is not worth
> >the fragmentation and other problems that would arise.  There are a
> >multitude of significant OO modules can could be used for this instead.
> 
> Well, you have nothing to worry about; I decided weeks ago to avoid 
> such problems and not write any modules named 'DBI' or 'DBD' for Perl 6.
> 
> Rather, the rewrite of 'Rosetta' in Perl 6 will satisfy the necessary 
> issues, both as an example of object oriented programming, and as a 
> working solution for native database access in Perl 6.
> 
> FYI, I am now positioning Rosetta as being a solution in the same 
> problem domain as DBI itself, where the main thing it "does" is 
> provide a generic lower-level transport layer between an application 
> and database (and is a foundation for more specialized APIs), and 
> that it is conceptually close to the database's native API.  The main 
> difference between Rosetta and DBI is that Rosetta does more munging 
> of what it transports, to further homogenize the appearance of 
> different databases.
> 
> See 
> http://svn.openfoundry.org/pugs/ext/Rosetta-Incubator/docs/Overview , 
> which I just wrote this monday, for the whole comparison; it is 
> incomplete, but readers should learn a lot about the intentions of 
> Rosetta that they didn't before.
> 
> Rosetta manifests my own ideas of an ideal "DBI v2", though I will 
> never name it 'DBI' to avoid stepping on your toes.  I won't give any 
> pretense that this is an official successor to DBI v1, though I'm 
> hoping that the official DBI v2 and my effort can cross-pollinate 
> good ideas.

That's great. Good luck.

Tim.