Re: Why J2ME? (was Re: [A5:RT] Minimum J2ME Specs and Strategy)

Stefano Mazzocchi wrote:

> Let me remind you that Avalon is a server side framework and it was 
> designed to be that way and the ASF created a project to substain that.

Avalon is simply a component framework--although it was originally
designed for the server.  COmponents can live right at home on a
constrained device.


> Can we make it client-side as well? sure, with a lot of painful rethinking.

I don't necessarily think so.  With the set of interfaces I proposed
here, we can easily wrap Avalon 4.1 functionality with an A5 face,
and vice-versa.  We also have an interface that works from J2ME on up.


> Should we? I'd say no, there are already enough irons in the fire.
> 
> Say you come up with back incompatible changes for A5 and say that you 
> go around the projects based on A4 telling them that you had to do those 
> changes to be compatible with J2ME.

It's not necessary.  A5 can be made primarily easier to develop with,
which is one compelling reason to upgrade.  THe fact that it can also
work with J2ME makes its usability go up.


> Do you *seriously* think they would follow your changes and modify 
> something that works perfectly as it is to match some elegance concepts 
> that apply to platforms they are not even interested in?

If A5 provides the tools to automate the upgrade path, *and* to work
directly with A4 components, then why is there such a big deal being
made?  The user does not lose their investment in A4 development, and
has a smooth upgrade path to A5.  A5 will contain many features that
developers will want.


> Please, people, thinking is nice and cool, but this is *NOT* an 
> accademic research project, it's a framework that must be *rock solid* 
> for several other communities to base their hard work on for years to come.

That is being kept in mind.


> *this* is way more important than any J2ME support, or you are asking 
> for a work.

Again, simplifying the API while providing legacy support is not only
doable it is vital.  We need to lower the boundaries to new developers
who want to use it, while not losing the investment that existing
developers already have.  It can be accomplished, and should be
accomplished.