Re: Re: Re: Have you looked at UIML?

On Friday 30 August 2002 12:24 pm, Gerald Bauer wrote:
> > > > Have anyone looked
> > > > at UIML?
> > >
> > >   Yes, I looked at UIML. It's a dead end and a
> >
> > piece
> >
> > > of junk.
> >
> > That only time can tell. But definitely not a piece
> > of junk!
>
>   How longer do you need to wait. UIML is more than
> three years old and isn't going anywhere. No book are
> forthcoming. No engines and so on. It's pratically
> dead already.

Sometimes it takes awhile for a technology to catch on. For instance, I've 
been using Python since 1994. I've taken alot of abuse from Java advocates 
for using an abscure language. Time is on my side, but it can be painful 
waiting. 

I'm not arguing that UIML is the right answer, but I do think it should be 
examined and reflected upon when making design decisions. It's like looking 
at the OSI model when talking about the IP stack. The work on the theory of 
the OSI stack was great, it was just too costly to implement all the options. 
IP, by constrast, came with working code. The failure of implementations of 
OSI to materialize did not invalidate the model. It turned out that the much 
slimmer IP specification was simply cheaper and easier to implement and use. 
The same model, but with a lower cost of ownership. (Also, the IP business 
approach of making source code freely available stomped the OSI business 
model.)

> > while UIML is built on a solid
> > basis of HCI research
> > with a sound specification.
>
>   Have you used UIML for any of your applications? Can
> you show me some links of live UIML apps? Or is this
> just a philosphical argument with no foundation in the
> real-world?

Let's not use number of apps implemented in the technology as a qualification  
for consideration. We'll all be using VB if that was the critical benchmark.

> > describe a device independent user interface with
> > regards to all of
> > its aspects.
>
>   That's the point. UIML is so abstract, you can use
> it for anything and claim that it's superior. All you
> have to do, is fill in the details.

By contrast the XUL definition could be called a hack by the Mozilla 
developers and that the hack is now being stretched to do things it wasn't 
originally designed to do. (This may not be the case, but if we are going to 
dis the other side we should be fair and do it in both directions.) 

I think XUL is great. I also think UIML has been well reasoned. Working code 
always trumps talk, but is XUL feature complete and does it scale? The 
Mozilla browser is certainly a good existance proof, so it probably does. 
However, is there a reason for not looking at the two of them? Let the 
details speak for themselves, but look at the details, don't argue about 
vague notions to proclaim one better than the other.

> > The time of UIML may have not yet come, however, it
> > just won't hurt
> > reading the UIML specification to get a faint idea
> > of what else waits
> > there to be considered...
>
>   Again UIML is years old and a case-study of how
> *not* to do it. Just using XML isn't enough.

The comparison between UIML and XUL is also a comparison between what happens 
when you have a well funded project with lots of highly skilled developers 
and when you have a minimally funded research project at a University with 
only semester long blocks of student time to work on the project. Sometime 
the available resources are more important to the outcome than the quality of 
the design.

BTW, I'm neutal on this issue, both have merit. I'd just like to see a more 
civil tone in the discussion. There's no reason to be harsh and jump down 
someone's throat for suggesting an alternative that is on target, albeit, a 
bit under-developed. 



-------------------------------------------------------
This sf.net email is sponsored by: OSDN - Tired of that same old
cell phone?  Get a new here for FREE!
https://www.inphonic.com/r.asp?r=sourceforge1&refcode1=vs3390