[Grok-dev] Re: five.grok and page templates

[Martin Aspeli]

Wichert Akkerman wrote:
> Previously Martin Aspeli wrote:
>> Hi,
>>
>> With a lot of help from Philipp, I made five.grok use 
>> Products.Five.browser.pagetemplatefile.ViewPageTemplateFile today. This 
>> is necessary to make it work with Plone, and I presume other code that 
>> makes "Zope 2" assumptions as well, e.g. the user of 'here' or reliance 
>> on Five's provider: expression support.
>>
>> We made this work by overriding the ITemplateFileFactory for the '.pt' 
>> extension from grokcore.view and implementing a ZopeTwoPageTemplate 
>> that extends PageTemplate and uses a ViewPageTemplateFile in 
>> setFromFilename(). We also overrode render() to do some aq wrapping and 
>> call the template in a different way.
> 
> Have you checked if this still works with z3c.jbot? That should be a
> requirement before using something like this in Plone.

I haven't.

I would definitely like it to work with z3c.jbot. However, z3c.jbot is 
neither a part of Plone core, nor Zope core, nor anything else, is 
fairly new, and does some deep monkey patching. I'm not comfortable 
recommending it as a long-term solution to people who are new to Plone 
yet, and I'm not comfortable committing Zope/Five/Plone to having to 
support it indefinitely.

This discussion belongs elsewhere, but if we want z3c.jbot to be such an 
official part of Plone that people take it into consideration when 
working on Plone (or Five!) code, then we need to submit it to the usual 
process of review and discussion and include it as a maintained, core 
package. We can't test with every third party solution out there. This 
isn't about quality or even ubiquity: It's about the processes we have 
established to help de-risk new approaches for those who do not 
regularly read lists like this and establish a stable, supported 
platform that doesn't grow willy-nilly.

I personally like the z3c.jbot philosophy very much. Overriding 
templates should be that easy. However, I'm not convinced that it is a 
strategically sensible implementation, and there has been no real review 
to prove me wrong, much as I'd love to be. In fact, I think 
grokcore.view holds more promise as a base for making customisation 
"that easy" in a less monkey patched way.

I will endeavour to test, though, and make adjustments to make z3c.jbot 
work if required and reasonable. I'm also interested in making z3c.jbot 
less hacky by proposing more robust changes to Zope (or Plone).

Martin

-- 
Author of `Professional Plone Development`, a book for developers who
want to work with Plone. See http://martinaspeli.net/plone-book