[Grok-dev] Re: Using martian to load schemata

[Martin Aspeli]

> Yes, you'd have a component called Schema that you'd grok for with a 
> ClassGrokker, and you'd define a directive called model. Something like:
> 
> class SchemaGrokker(martian.ClassGrokker):
>      martian.component(Schema)
>      martian.directive(model)
> 
>      def execute(self, class_, config, model, **kw):
>           ... register your action here...
> 
> You'd need to use at least grokcore.component to get the 'grok' 
> directive available. Martian itself doesn't interface with 
> zope.configuration.

I'm happy to depend on grokcore.component. In fact, I may start using it
for other config, and this would be a nice way to sneak it into use with
Plone. :)

> No, that's unrelated. The 'config' parameter comes from the 
> zope.configuration machinery. It's passed in to the 'grok' ZCML 
> directive implementation as implemented in grokcore.component.zcml.

Aha, I see. So it's passed around in kwargs from that point onwards?

> > I'm also not quite sure how to use config.action() to ensure that my 
> > particular callable gets executed after all standard components have been set
> > up, though I guess that's a zope.configuration question?
> 
> Give it an 'order' argument.

Thanks a lot! I'm going to try this out tonight, hopefully. It sounds kind
of fun. :)

Cheers,
Martin