Am 08.04.2010, 19:33 Uhr, schrieb Iain Duncan <iainduncanli...@gmail.com>:
> Cool. that makes sense. So here's a question that didn't seem immediately
> obvious from the docs and could possibly be made clearer as a foot note
> there somewhere ( or maybe I just missed it ), can context be used with
> interface? I totally understand the motivation to not force people to
> understand or care about the ZCA and interfaces, but again, a big
> for me in switching to bfg is the ZCA integration, so *maybe* it could be
> mentioned more obviously that we can do view look up by interface as
> well as
> class name?
I think the ability to use the ZCA is one of BFG's secret weapons - for
small, single issue sites you're hardly likely to care so it shouldn't get
in your way and BFG let's you get up and running - but anything that's
likely to grow will, will not get very far without it.
Regarding the semantics: the ZCA use of for is slightly misleading as
views are multiadapters (of the request and the object to be published).
Actually they are the most difficult aspect of adapters to explain - when
using a mask we are likely to gloss over the role observer but without
whom the mask has no function. Note to self, try and come up with really
metaphors that are really easy to understand.
> It's just a gut impression I had reading the new docs, they are excellent
> and super clear, but I feel like it's harder to make the connection with
> the under-the-hood Zope based principals than it could be.
Yes, no need to apologise or fear calls of "eek, Zope is complicated!" -
despite the size of the eco-system there are new people coming to Python
everyday who've never heard of it. But this best done by providing more
useful examples of when this is appropriate. Maybe the cookbook
application, which relies on views "for" specific interfaces, could serve
Clark Consulting & Research
Repoze-dev mailing list