[Zope3-dev] XML Schema support on Zope3

[Gary Poster]

On Aug 16, 2005, at 7:31 AM, Julien Anguenot wrote:

> Gary Poster wrote:
>>> FWIW, I looked at this a couple of years ago and concluded that,   
>>> with
>>> a reasonable amount of work, zope interfaces could support all  of
>>> XMLSchema, but not vice versa.
>
> Sure, but if necessary we could add a 'zope' namespace for power users
> who want to take advantage of all the zope.interface (schema) features
> from XML but we need to be compliant with the standard XML Schema
> specification so that it will not be compulsory to have any zope
> knowledge to built zope compatible schemas from XML standard tools.

Right.  However, I think (in fact, was convinced by a third party  
years ago) that the use of such a namespace will likely make using  
the XMLSchema in a automated schema editor as annoying as ZPT in  
Dreamweaver: possible, but likely prohibitively annoying.  If  
roundtrips are desirable, users will be better served to constrain  
themselves to the capabilities of XMLSchema.

>> BTW, the obvious following question to my email is "how are they
>> insufficient"?  I did this two years ago, and sadly don't  
>> remember  the
>> details, so feel free to ignore me as without proof.
>
> I don't want to ignore you on this :) If you got pointers on the work
> you did at this time it would be really appreciated :)

No, that was three computers ago.  Long discarded. :-/

> [...]
>
>
>> So again, not intended to be throwing cold water, just  
>> observing.   For
>> some use cases, such as making a system designed to be customized  by
>> minimally-trained users (yours, AIUI), the advantages of  
>> XMLSchema  are
>> likely more important than its disadvantages.
>
> The point is mainly to be able to support XForms and as well, as you
> mentionned, being able to let the framework "users" (integrators for
> instance) define their schemas with tool and a technology they know
> already...
>
> Note, that after the XML schema is set on the interface it's  
> possible to
>  work and extend it as it was defined in "pure" Python so you get both
> technology advantages.

I did notice that, and that is cool.

> Got time to help me on the implementation ? ;)

:-)  Good question, but no, that's not in the cards right now.

Gary