[Product-Developers] Re: Where does it hurt?

[Daniel Nouri]

Michael Hierweck writes:

> Dylan Jay wrote:
>> Daniel Nouri wrote:
>>> Michael Hierweck writes:
>>>
>>>> Reinout van Rees wrote:
>>>>> Martin Aspeli schreef:
>>>>>
>>>>>>  - Areas where there appears to be more than one approach, and it's
>>>>>> not clear which one to choose
>>>>> Hooking templates up with views. I scream and curse every time someone
>>>>> puts a hardcoded ZopePageTemplateFile('myhardcodedname.pt') in code. I
>>>>> had to *monkeypatch* plone.app.content to wire in a different template
>>>>> to render the folder listing (order of columns had to be different).
>>>> Up to now I thought statements like
>>>>
>>>> render = ViewPageTemplateFile("xyz.pt")
>>>> __call__ = ViewPageTemplateFile("xyz.pt")
>>>>
>>>> as if found them in Plone 3 for Viewlets and Portlets and other
>>>> resources were good, modern style and best practice.
>>>>
>>>>> If we're going to be buried in zcml, please also use it for the places
>>>>> where it is actually useful. Like the 'template' attribute so that you
>>>>> can override it.
>>>> How to make it better/more configurable?
>>>
>>> If you're using your own layer, you can just register a view with the
>>> same name for your layer.
>>>
>>> You can't actually customize a view with the template argument unless
>>> you use overrides.zcml or register it for something different.  So it
>>> doesn't make too much of a difference if you assign the template in
>>> Python or ZCML.
>>>
>>> Of course, there's valid uses of registering a view class with a
>>> different template.  To allow this, the author must use the 'index'
>>> class variable name for the template, like so:
>>>
>>>   >>> class MyView(BrowserView):
>>>   ...     index = ViewPageTemplateFile("xyz.pt")
>>>
>>> This allows you to register the same class with a different template,
>>> for a different name or context.  (Naturally, registering it for the
>>> same name and context will fail, unless you use a different layer.)
>> 
>> great example of 2 different ways to do the same thing but have
>> differetn unforseen consequences. 
>
> Sorry, I did not understand the second, more flexible possibility.

The idea is that, if you use 'index' as the class attribute name for
your template, you'll be able to override it *even if* the template is
assigned in Python code.  This is equivalent to specifying the template
via ZCML.