Re: zope 3 schema's and the rest of us

On Thu, 13 Jul 2006 20:56:20 +0200, Roché Compaan wrote:
> On Thu, 2006-07-13 at 14:57 -0230, Rocky Burt wrote:
>> On Thu, 2006-13-07 at 19:20 +0200, Roché Compaan wrote:
>> > The fact that UI definition is so tightly coupled with AT Schema
>> > definition is bad design in my opinion (even though it may be
>> > convenient at times).
>> 
>> +1 here.  Separation of concerns is a good thing and has been proven as
>> such over and over again.
>> 
>> > I don't necessarily want views generated from AT schema's. What I
>> > would like, is that I get the same amount of boilerplate views
>> > generated from a z3 schema as with the current AT. I'll do the rest an
>> > convert the AT schema to a z3 schema. That and references would be
>> > bliss ;-)

i don't necessarily want views generated from AT schemas either, but i do
sometimes want to be able to extend existing AT objects in a Z3 way
WITHOUT having, at that moment, to port the entire object over to a Z3
schema.   no matter how well we do this, there will certainly be wrinkles
of behaviour differences that come up when we start porting our AT schemas
to Z3 schemas, and there are gonna be times when a deadline-driven
developer will have to punt on doing a full port.  this is the driving use
case, in my mind, for providing a smooth story for allowing both AT- and
Z3-based schemas to exist and to generate UI on the same object. 

> In essence what I am saying is, rather than trying to "adapt" AT to z3 I
> would rather port what works well in AT to z3 and then drop AT. I would,
> very simply:
> 
> 1. Port missing fields
> 2. Port missing widgets
> 3. Rewrite content types
> 
> It's only 3 steps, it should work ;-)

in a perfect world, i'd agree with you.  but consider the following two
situations:

1. there's no clearly defined way to mingle AT and Z3 schemas on the same
object:

developer foo is tasked with adding some new features to an existing
project.  he's been following the lists and lurking on IRC, and he knows
that Z3 is the way of the future, so he decides that he's going to use Z3
tech to do so.  he reads the formlib tutorial, he learns about Z3 schemas,
and he gets to work.  his only option is to rewrite all of the
pertinent content types using a Z3 schema, and then migrate his data from
the old content to the new content.  but somewhere along the way he hits a
wrinkle.  he spends a day working on it, but can't quite get stuff working
the way he wants to, so he punts, goes back to the tried-and-true AT way,
and gets the job done in 30 minutes.

2. there IS a clearly defined way to mingle AT and Z3 schemas on the same
object:

i hardly feel like i need to spell this out, but here goes:  developer foo
has the same problem in scenario 1.  instead of having to rewrite his
entire content type, he has an easy way to add Z3-based behaviour to his
type w/o sacrificing the existing AT-based functionality.  he spends
all of his cycles working on just the new behaviour.  it takes him longer
than it would have w/ AT (just due to familiarity issues), but it's still
no more than half a day.  he feels good about his success, and decides to
use Z3-based schemas for ALL the new functionality he builds from now on.

i'm strongly in favor of story number 2, myself.

-r




-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642