Re: How to approach converting to CDBI?

[prev] [thread] [next] [Date index for 2005/07/04]

On Mon, Jul 04, 2005 at 01:36:38PM -0400, Brett Sanger wrote:
> On Mon, Jul 04, 2005 at 06:27:24PM +0100, Matt S Trout wrote:
> > Not at all. Use Class::DBI::Sweet as your base class instead of Class::DBI
> > and you can do
> 
> Isn't that at a 0.3 release?  It looked interesting, but while my
> sysadmin understands that tendancy of free/open source projects to have
> version numbers that asymptotically approach 1.0, I can't see him being
> happy with a 0.3, particularly when that accurately depicts a young
> code-base.

However, it's a young code-base which passes the Class::DBI tests itself
3 times before it says "yes, I'm ok to be installed" :)

The current Sweet code-base is theoretically young, but the code in it is
largely wheels all the Sweet developers have invented at least once before
- just now collected in one place so we have a feature set more comparable
to ActiveRecord (and in some areas substantially better).

> Let me switch over to CDBI and get CDS a few months to settle, and that
> might work.

You'll find that if you don't set a cache, or call the prefetch code, or
etc. etc. then Sweet will actually make very minimal differences to your
execution path over base Class::DBI - I'm hoping to separate out each of
the feature groups into a separate module so you can load them one at a time
but currently what you don't use generally gets short-circuited past, so
you'll be using very little "new" code.

Sweet is already in production at a number of our clients in order to beef
up the capabilities of our Maypole apps while I find time to port the
appropriate code-generators to Catalyst ...

Of course, none of this stops it being a 0.03 release, but I thought maybe
I should clarify a little :)

        -- 
             Matt S Trout           Website: http://www.shadowcatsystems.co.uk
  Technical Director        E-mail:  mst (at) shadowcatsystems.co.uk
Shadowcat Systems Ltd.