[Grok-dev] Re: RFC: Masterpage implementation without macros

[Philipp von Weitershausen]

Martijn Faassen wrote:
> Philipp von Weitershausen wrote:
>> Martijn Faassen wrote:
>>> We need to try some implementation work. Either lxml or Genshi should 
>>> offer the tools to implement the xpath stuff. It also needs work to 
>>> define what grok.skin() actually does and then to implement it.
>>
>> I'm much in favour of having an XML-based pipeline (both lxml and 
>> Genshi sound like they provide good tools here).
>>
>> Gary Poster just proposed (see my response [1]) something on the 
>> zope3-dev mailinglist to allow standard browser views to return 
>> non-string objects which the publisher then adapts to IResult in order 
>> to actually obtain a string. 
> 
> Just proposed? This:
> 
>  > [1] http://mail.zope.org/pipermail/zope3-dev/2005-December/016796.html
> 
> Goes back to december 2005. :)

I phrased that wrongly. He was proposing to turn an implementation 
detail of the publisher regarding the streaming of large data into a 
general hook for view result transformation.

> Anyway, allowing the standard browser views to return something that 
> needs to be adapted to IResult would allow us to return markup streams 
> that could then possibly be post-processed. If we can at least hook into 
> the right place of the publisher to do the post-processing?

That's what the IResult adapter is for. It multi-adapts 
(return_value_of_a_view, request) and returns an iterable of text. If 
return_value_of_a_view were an lxml tree, you could

* modify the tree (e.g. using Python API, XSLT, etc.)

* serialize the tree to a string

in that IResult adapter. Naturally, we can do the same thing with 
different technologies (Genshi comes to mind).

> Meanwhile it looks like Kevin Smith's skin work on his branch can just 
> go ahead, as we'll need it whatever we do. We should look into merging 
> that.

Agreed.


-- 
http://worldcookery.com -- Professional Zope documentation and training