Re: Modifying data models within the template

Friday, July 8, 2011, 3:23:42 AM, Chris wrote:

> I implemented my own ObjectWrapper and that seems to work.
>
> Question #1 is still open, though. Does anyone know if modifying a data
> model during template processing will lead to threading problems?

I'm just answering really quickly and without reading much of the
thread... If multiple threads share the same POJO-s then it will cause
problems unless the POJO-s themselves are thread-safe and you only
share the POJO-s, not the wrappers themselves. If you just change the
model within a single thread, use vanilla BeansWrapper, that meant to
handle changes. (I pretty much never use DefaultObjectWrapper...)

As of templates with different data-models... maybe it's not
applicable to your case, but when I have used FreeMarker to generate
"components" on a page, each component was an independent FreeMarker
template with its own data-model. That I did by simply calling
Template.process for each components. And yes, some of them was nested
into other components; that should work.

-- 
Best regards,
 Daniel Dekany

> On 7/7/2011 4:29 PM, Chris wrote:
>> A bit of poking around shows that the default ObjectWrapper wraps Maps
>> in a SimpleHash, and the SimpleHash copies the map. That's the answer to
>> question #2. Is there an easy way to get it to stop doing this?
>>
>> Making a copy of every hash in the system strikes me as an unnecessary
>> performance killer.
>>
>>
>>
>>
>> On 7/7/2011 4:18 PM, Chris wrote:
>>> We need to give [#includes] and subpages their own datamodels to keep
>>> things modular. The scheme I've come up with is to create a directive
>>> that can be included at the top of each template that will add that
>>> template's data to the top-level data model.
>>>
>>> Two questions:
>>>
>>> 1. This is an explicit violation of the "data models are read-only" rule
>>> described here:
>>>
>>> http://freemarker.sourceforge.net/docs/pgui_misc_multithreading.html
>>>
>>> If a new datamodel is created for each page fetch, is this a problem?
>>> I'm running in a servlet environment. So far as I can tell, only the
>>> thread that is processing the page will ever see this data model.
>>>
>>> 2. When I use a HashMap as the main "root" data model, the scheme fails.
>>> The template says that the variables I've added to the model do not
>>> exist. The scheme *works* when I have the root HashMap point to an
>>> object, and then I replace the value of that object. There's something
>>> about adding a new key to the HashMap during template processing that
>>> the system doesn't like.
>>>
>>> Is the template processing code playing games with my HashMap, like
>>> making a copy? Why?


------------------------------------------------------------------------------
All of the data generated in your IT infrastructure is seriously valuable.
Why? It contains a definitive record of application performance, security 
threats, fraudulent activity, and more. Splunk takes this data and makes 
sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-d2d-c2