comments inline ...
Dave Johnson wrote:
comments inline ...
On 6/21/06, Allen Gilliland <[EMAIL PROTECTED]> wrote:
David M Johnson wrote:
> On Jun 20, 2006, at 9:05 PM, Allen Gilliland wrote:
>> how come we need WeblogEntryPluginsHelper? i thought we were going
>> with overriding pojo methods?
>
> Yes, I'm not happy about the current situation.
>
> The problem is that:
> 1) Some plugins need to do some weblog specific setup, to read the
> bookmark collection for example. To optimize performance, it's best to
> init the plugins upfront. If that cannot be done upfront, then it must
> be done each time entry text is to be rendered.
Yes, I agree that kinda sucks, but we've lived with that up until now.
I don't know how to fix that either, but I think we can continue to live
with it and figure out an optimization later.
We've got the optimization in place now with 2.0. Plugins are
initialiazed upfront.
Huh? How? My understanding is that currently the PageHelper class
initializes the plugins when it is instantiated and the PageHelper is
constructed once each time the ContextLoader.setupContext() is called.
It doesn't make sense to me that we could have plugins initialized
upfront in 2.0 because the problem is still the same. Each set of
plugins has to be customized per weblog.
> For #2, I think we should change plugins so that they do NOT add things
> to the context , since the mechanism for adding things to context is to
> use page models.
I agree, the only plugin that I know this affects is the TextilePlugin
and it doesn't look like it will make the plugin break.
The Wiki plugin, one of my faves, will be broken. But I will fix it to
support 3.0 by using the page model approach.
Ah, yeah, that does suck. That one's not in the Roller codebase so I
didn't know about it.
What is the plan for fixing it via page models? Not have it considered
a plugin, instead have it be a custom page model?
-- Allen
I think that it's better for us to try and make this change sooner
rather than later. We know it's a change that needs to be made, so
eventually we are just going to have to bite the bullet and do it. Why
not now?
Yes, I agree. Templates will be written against this stuff, so we need
to get it right now.
I'm going with POJOs methods (e.g. getTransformedText,
getTransformedSummary) as I originally proposed -- and try to figure
out how to get some form of caching going. I can't accept walking the
bookmark collection everytime somebody calls getTransformedText.
- Dave
