[Zope-CMF] Re: Directory-based skin layers?

[Ricardo Newbery]

At 1:35 AM +0200 4/23/04, Dieter Maurer wrote:
>Ricardo Newbery wrote at 2004-4-21 16:43 -0700:
>>Again, I'm trying to grok how this skinning machinery works and
>>looking for the best way to insert "local" skin layers (as opposed to
>>the "global" layers in the skins_portal).
>
>Not sure, I understand what you mean with "local".
>
>When "local" means "inside a subhierarchy",
>I would use another "Skinnable" object as root of this
>subhierarchy and define the "local" layers on this object.
>
>--
>Dieter


Sorry, I guess I should have included the text from my first post.

Basically, I would like to enable directory-based skin layers which 
would take higher precedence than the layers defined in portal_skins. 
I'm NOT talking about an access rule that would reset the skin name 
OR a complete override of the portal_skins layers but rather a 
mechanism that would acquire any skin layer folders in the current 
directory and its parents and prepend them to the portal_skins layer 
list.

I'm calling these directory-based skin layers, "local" skin layers, 
and the ones defined within portal_skins, "global" skin layers.

I think I've puzzled out how to do this except for one part.  A 
Skinnable object first acquires from portal_skins before performing 
the usual acquisition lookup.  If I can get my modified portal_skins 
tool to collect the local skin layers (a subclass of SkinsContainer?) 
and prepend them to the global skin layers and return the result (as 
a chain of acquisition-wrapped folders), the rest should be easy ;-)

The problem is I'm not sure how to get the portal_skins tool to 
collect the local skin layers.  As far as I can tell, the tool 
doesn't "know" anything about the original object that acquired the 
skin tool so how can it acquire the local skins that exist in the 
original object's acquisition space?  Hmm... something just occurred 
to me...  Since the skin tool is acquired via getattr, maybe I can 
just crawl back up to the original object using aq_parent (can I do 
this?).

Or is the better solution just to patch SkinnableObjectManager (maybe 
just adding a couple of lines to __getattr__ to acquire the local 
skin layers before acquiring the global skin layers) and leave the 
portal_skins object alone?

Thanks for the help,
Ric