[Seaside] Pondering if Dojo javascript toolkit is more convenient than Scriptaculous for Seaside solutions

[Sebastian Sastre]

Hi there,
 
    I like to share some ideas about the value equation (scoring) of Seaside
systems (solutions) and to purpouse to evaluate somethig that I suspect that
could add value this equation so, if this suspiction is finally right, it
will add value for all the comunity, so we all won.
 
    Imagine you have a scoring equation wich defines the value of your
seasside solution. This equation have the general form of:  X1+X2+..+Xn 
 
    Each term is a function of value (scoring) of different domains: for
instance.. X1 is loadbalance, X2 is memory scalability, X3 is processor
scalability, X4 is complexity scalability, etc. As the terms can be complex
scoring functions, this equation is also a sort of composite if we'll
express it in "patternese".
 
    Given that, I want to make focus on two terms that are important for
developers: 1. Content and 2. Form. 
 
    For (non static) web applications (or online systems as I like to say)
Seaside is a framework that irrefutably takes care of "the contents" and
solves basically and consistently "the form" using html and CSS. In the form
term I want to ponder a (sub)equation of two terms:
 
    1. Usability. The digital ergonomy. The empirical ease of use of the
systems. The facilities that your applications had to offer to end (daily?)
users.
    2. Cosmetic. The indumentary of the systems. Pure eye candy makeup which
users like to, as it's clothes, be able to renew often and flexibly and
numerously.
 
    For the contents I have no further comments because I consider Seaside
to be vanguardist and by far the best in it's field. So for me, content has
reached irrefutability. This means that for the "sale" (final score value)
this term sells OK. Now.. for the form (usability+cosmetic) we have reached
an acceptable scoring but I think I found a way in wich we can improve those
terms in a way in direction of form (usability+cosmetic) irrefutable scores.
 
    Said that, the concrete purpouse is to evaluate Dojo javascript toolkit
as being used for Seaside applications. (see: http://dojotoolkit.org/)
 
    I read that Dojo is ready to be use from Ruby so I think that nothing
should stop us to use it also from Seaside. Seems to be that Dojo community
has some support from IBM so this should weight something in the balance. A
first look into the demos could show you nice form value. If you take a look
on the widgets in the API documentation, try to forget cosmetic for a moment
and see the usability that those widgets have to offer (and I mean the
contents and not how they made the documentation interface). Take a look
into the panes with draggeable splitters for instance. That can't be done
with css. CSS can't even give us (yet) an appropiated general solution for
elemental #north, #south, #center automatic layouts. To make that you end up
using CSS hacks that leads you to have an early compromise with the page
shape. This deteriorates the "desktopability" of web applications.
 
    I don't know you but I found those widgets very good in what they do and
I think that to have them in sum to seaside applications will rise the score
of seaside solutions significantly.
 
    My experience shows that wrapping this toolkit should lead us to a
hierarchy as rich as any Smalltalk View hierarchy of a Smalltalk that is
using native widgets. So this means coupling because the classes end up
wrapping system events and calls to build the views. Here there will be Dojo
events and Dojo calls to build up the windows (yes it can do windows inside
a page) in the browser. But the add of coupling will be with a
cross-platform javascript widgets framework so we won vanguardist widgets
smalltalkish way usable that work for lots of browsers in lots of OS's. So
with this choice we end up using javascript calls as the dlls (or
homologous) libraries calls of a native system (which is the browser in this
case).
 
    Finally, if a Seaside is about the "descktopability" of web
applications, then Seaside+Dojo could be a good deal. Once the wrapping it's
done and maintained, It will modify the whole equation allowing the whole
score to raise more dramatically for the same development effort. That will
maintain Seaside and Smalltalk itself to the vanguard allowing the creation
of Smalltalk solutions of high real value. An ammount of value that other
web application frameworks could only dream about.
 
    Now the realistic questions: 
 
        Why do you  or do you not found Dojo a convenient choice?
        Beside Scriptaculous, do you know another choice than Dojo? it will
lead us to where? how far?
        Do you admit that wrapping Dojo will lead us to a View hierarchy?
        Admiting that this frameworks are a View...
            Do you found Model-View-Controller can be used in Seaside
applications?
            Do you found Model-View-Presenter more suitable?
            Do you found another event-model-view glue more convenient than
those? which one? why?
 
    cheers,
 
Sebastian Sastre

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/seaside/attachments/20070426/306d847a/attachment.htm