On Fri, Mar 22, 2002 at 02:59:00PM -0800, Jeff Zucker wrote:
> Tim Bunce wrote:
> 
> > And it's not good I'm afraid. You've effectively hardwired a specific
> > driver in place of the DBI.
> 
> What I sent you is a proof of concept, not a final product (tho it works
> for its one driver).
>
> > It only emulates the DBI in the narrowest
> > sense. You can't use it to load and run any random pure-perl driver.
> 
> I wanted to hear from you whether this project is even worth doing
> before implementing that.  I'd previously done some preliminary tests
> with DBD::XBase and it seems to work well.  Obviously it can work with
> DBD::CSV and DBD::Excel and any other drivers that use SQL::Statement. 
> I totally agree it should work for all pure-perl drivers.
> 
> > And that's a pity because there's (probably) no big technical issues
> > in creating a true pure-perl DBI with a reasonable level of functionality
> > (but much slower performance).
> 
> Is this a project worth embarking on?

Maybe :)  It would certainly rate as interesting and challenging.

> If so it would proabably be best
> as a team effort pooling resources of all the pure perl driver authors
> and whomever else might want to contribute.

Just to clarify: It need not involve the pure perl driver authors
because no driver changes should be required - but they may have a
natural vested interest.

[EMAIL PROTECTED] is the place to talk about it.

> > DBI::PurePerl would start with 'package DBI;' and define all
> > the functions and methods that DBI.xs currently does.
> 
> My previous thinking was more along the lines of providing only the
> basics so that no one would be tempted to use it when they really should
> be switching to DBI.

I think the performance drop will take care of that :)

But there's no point in limiting yourself. Once you've a working
method dispatcher you've instantly got ~90% of the DBI working.
That's the way to go.

Tim.

Reply via email to