[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

Re: The Browser Wars are Dead! Long Live the Browser Wars!

• To: mbatsis@n...
• Subject: Re: The Browser Wars are Dead! Long Live the Browser Wars!
• From: tblanchard@m...
• Date: Mon, 21 Oct 2002 15:57:46 +0200
• Cc: xml-dev@l...
• In-reply-to: <3DB3F0D9.30404@n...>

On Monday, October 21, 2002, at 02:19  PM, m batsis wrote:
>
> This [1] has been my favorite applet for some time.

Ick.

>> I looked at UIML and it looks like a bad joke to me.  I can't 
>> remember when I've seen something less readable and more complicated 
>> that did so little.  So much noise for so little signal.
>
> So little? Reusing UI presentation and logic, including behaviour 
> hooks for interaction logic to output to different UI formats such as 
> Swing and HTML is lots more that "so little" and I have not seen any 
> contender to UIML yet.

Maybe because nobody else thinks its a good idea? I don't see it.  Nor 
do I believe in the underlying goal - self adaptable UI's aren't 
practical.  They are always very clunky.  You need to rethink task flow 
completely when shifting devices with different constraints.

>> Your later comment about hammers is particularly applicable.  ML's 
>> are really poor mechanisms for describing behavior.  They're sort of 
>> poor mechanisms for describing relationships (they impose a sort of 
>> directional view via the element nesting that is artificial - is 
>> artist inside of CD or is CD inside of artist - depends).
>
> XML is a format with hierarchical relationship semantics build-in 
> while non-hierarchical relationships can be described easily. And it's 
> a text format, not a proprietery representation you cannot send over 
> to your partner without tons of documentation and code for him to be 
> able to use it.
>

Its heavily biased towards hierarchical. I don't think the second 
statement is a sentence.

> Verbose and unreadable? It's by far the more efficient mechanism I 
> have seen in action,
>  it works without compiling enything and it's possibilities are 
> unlimited.

OK, nothing fits this scenario.  All possibilities are limited.  As for 
verbose and unreadable, I was referring to the UIML thing.  It is, if 
you'll excuse the expression, clearly a cluster [expletive deleted].  Sadly, the thing 
is in PDF so I can't copy the example to show you, but it takes 
something like 10 lines to say

element setTarget: aTarget action: #performClick:

In fact it took me several minutes to penetrate the cruft to figure out 
that this is all it did.

OTOH, there is perhaps a certain usefulness in the XBL thing.  If I 
read it correctly, its primary purpose is to pack together bits of 
JavaScript into components.  I like that.  Using WebObjects, I could 
build components that render that way (I like WebObjects design 
paradigm - it builds a conventional GUI on the server that happens to 
render by emitting HTML and it delivers page requests as UI events - 
this is sensible but limiting in some ways as the browser's event 
delivery rate is not too good).

> It also utilizes browser capabilities to the maximum.

Is it supported in Explorer?  Because maybe I'd use it.  But even this 
tiny bit of improvement (for me - its just a tweak in the emitters in 
the WebObjects UI elements) doesn't address the underlying issues of 
binding UI to business models in a clean reusable way.

Todd Blanchard
System Architect

• References:
  • Re: The Browser Wars are Dead! Long Live the Browser Wars!
    • From: "m batsis" <mbatsis@n...>

• Prev by Date: Re: Perceived C/B for XML 1.1
• Next by Date: Re: The Browser Wars are Dead! Long Live the Browser Wars!
• Previous by thread: Re: The Browser Wars are Dead! Long Live the Browser Wars!
• Next by thread: Re: The Browser Wars are Dead! Long Live the Browser Wars!
• Index(es):
  • Date
  • Thread