Re: Four ways forward for MVC

Magnus Lie Hetland wrote:
> Arf!
> 
> I just wrote a veeery long email on this subject, and then my machine
> crashed. Oh, well... Here we go again (extremely shortened version).
> 
> It seems that we need to resolve the MVC issue before moving on to
> 0.2a0. From our discussion, I've gleaned four main approaches...
> 
> 1. Keeping the original 0.2 MVC mechanism
> 
> 2. Using the registerFooModel mechanism
> 
> 3. Using the event system
> 
> 4. Dropping MVC
> 
> In my original email I described all of these with pros/cons -- right
> now I'm too frustrated (due to my computer crash) to rewrite it...
> Maybe I'll do that later ;)
> 
> Anyway... There _are_ advantages and disadvantages to each of these.
> Without deciding anything here and now, I'd just like to temper the
> "drop MVC" approach slightly: Even if we drop all built-in MVC
> support, we can still have a 'modified' event (dormant until used by
> link()) that can then be used to implement MVC outside the
> proxy/wrapper architecture. And... If we do drop MVC in this manner,
> is there any need for in-place modification anymore? And if not, what
> should be the alternative? Copy semantics? Only immutable values
> allowed?
> 

I didn't follow all of your discussion but I would like to point you to an 
interesting article series written by Allan Holub: 
http://www.javaworld.com/javaworld/jw-07-1999/jw-07-toolbox.html
He discusses MVC and introduces an alternative approach he calls "Visual Proxy 
Architecture". Although all examples are written in Java, the principles do not 
depend on it and may easily be adopted for Python.

Michael

P.S.: Forgive me if I'm carrying coal to Newcastle :-)



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf