Oh NOOOOOO ... sigh.

> I've seen some examples that show how to set that up, but it seems like a 
> pain. And, if it's a pain for me, it's going to be a pain for anyone who 
> wants to use our proposed JNLP-based Preferences API.
I understand your trouble, and i agree with you.
But if we wouldn't use this class from Pivot classes, leaving it for
application developers ?

> - JNLP seems like a pain to work with
yes, it has always been, but it's mainly a setup problem.
For the Debug, yes this is the more complicated thing but again it's a
setup problem.

> - We could base it on the file system, but that would required signed applets
Or a normal standalone application (deployed with a zip file for example).
My first implementation of Preferences for Pivot was using a base
class and 1 implementation for File Systems (signed applet or
standalone application signed or not, saving data by default into user
home) and another implementation for Web Start Persistence, leaving
the choice to the developer on what to use, in a transparent way.
Was a little more complex than the current version, and should be
revised for Tasks, usage, etc. if needed.

> I'm leaning towards bagging Preferences for now (at least, until Sun makes it 
> easier to launch and debug JNLP-based applications). For now, I'd suggest 
> that apps that need to manage user preferences can simply do it with custom 
> code. JSONSerializer provides most of what you'd need to manipulate 
> JSON-based preferences, anyways. This way, it's up to the app to decide 
> where/how they should be stored.
Ok, but at least a common skeleton (generic, not tied with any backend
implementation) could help ... to not reinvent the wheel any time.
I understand that this is not a Pivot focus, but i think that a modern
RIA toolkit like this should give a sort of Local Storage facility to
being able for example to cache data.

> Thoughts?
At least, instead of throwing away the code, if we move it to a new
experimental subproject (not used by other pivot subprojects),
generating its jar ? And see what to do later, after someone has used
it (at least me) for some time ...


I have no problem to finish this part of code, but at least if someone
wants/uses it, otherwise it's time wasted. Tell me what do you think.

Thanks,
Sandro

Reply via email to