The Disappearing Desktop - It's Real This Time

[Andre Garzia]

Dan,

I was going to comment this on your blog, but I fell this will be a  
big email and I know blog comment engines are not suited for this (by  
the way, where is your RSS feed?).

I read the memos and I follow slashdot, osnews and a couple other  
sites with religious practice, I also am a subscriber for Doctor  
Dobbs Journal for I fell that before anything happens in the world of  
computing, they will shout. I also know that people here are much  
more experienced than I so my takes might sound naive. Well, enough  
off topic talking, let us get on topic.

We've be hearing this for a while, which does not make it true or  
false. The mais question I'd ask is: "Is the Web ready to hijack the  
desktop?". We live in a world of almost ubiquitous networking but  
there are times when one is not network, those times the web cannot  
handle. What if you need to use word processor on a vacation where no  
net access is available. For this kind of task, the desktop will  
always be there, even if the desktop moves toward being a single- 
machine-network. But if we think we're talking about a always present  
network with large enough broadband to make the packed juggling fast  
we need to think about some other issues.

User Interface. Browsers were created to display hypertext. All the  
rest was hacked inside. I know xul can create complex user  
interfaces, but we must remember that using AJAX and that means using  
CSS and HTML to construct user interfaces is not the easiest thing on  
earth. One thing is to worry about linux, windows and macs, other  
thing is to worry about each version of safari, mozilla, firefox for  
each platform, this and we're not even entereing the amazing realm of  
screen resolution and what does pesky little Firefox extensions can  
do with rendering of your website. Meaning it's harder to create a  
user interface for web based apps than it is to create for desktop  
apps. Next problem is session management.

The web was never created to be a place where you needed complex  
session management. Cookies were the simplest thing ever and they  
work fine for simple web stuff like shopping cart and user  
autentication. Now think of all things a desktop app needs to track  
and thing of the stateless nature of the web. Unless people invent  
some new way to store runtime information with a web browser, there  
will be no sane way to track runtime info. If you going to use a mix  
of server stored vars and javascript engine vars in memory then allow  
me to drive you to the most fricking thing ever to appear on your  
AJAX neighbourhood, BOOKMARKS!

We all know about the asynchronous nature of AJAX apps. Keep  
everyhing on the client machine, do asynchronous trips to the server  
when you need info from it. What happens if the user Bookmarks? After  
fiddling with the app, the user goes there and press "add bookmark".  
This add bookmark is not like a "freeze state" from emulators, when  
you try to recall that URL, the things that were on memory will not  
be anymore, this alone can break most AJAX apps. Not only bookmark  
but the most simple go back and go forward of browsers will wreck  
most web apps. Try fiddling with that while gMail is thinking, you'll  
see it getting lost pretty soon.

Meaning, it's hard to present a nice UI, it's hard to track runtime  
data and it's not user friendly, the nature of bookmarks, go forward  
and go back makes the browser experience of AJAX app very bug- 
friendly. I'll not dive into the beauty of Javascript as an app  
development language mixed with HTML, CSS and XML. Note that I havn't  
said things are not possible, but that they are hard to the point of  
being annoying and bug-friendly. That's why I don't think we'll be  
leaving the desktop soon, at least, not until someone present a  
better language than javascript, a better way to track and save  
runtime data than cookies and a better user interface than "hack your  
own".

Now where we stand? We have the dreamcard player which is smaller  
than most browser downloads and solves all kinds of problems quoted  
above. Revolution stacks can handle vars in a sane way, can provide  
any UI we need and can save/restore data easily. And they can be  
loaded on demand from the network. We have the technology everyone is  
looking for and yet, we're also lost. The problem we face is like the  
chicken and egg problem, third party cannot execute rev code if they  
don't download the player, if they're going to download the player  
they might just download a standalone. If we could make the player, a  
plugin and make the browser fetch it if needed, things would be  
easier... The problem is that since everyone and his dog got a  
browser installed, we're trying to hack apps into browser technology  
instead of trying to make a breakthru on network app tech.

We have the dreamcard player which is the most advanced and friendly  
thing ever. Before the dreamcard player, I used REBOL and REBOL had a  
browser plugin which was nice, there was also another solution called  
CURL, even Squeak could load modules on demand and be very user  
friendly. Even OpenDoc could solve such things. AJAX is not the  
answer for me. We need smarter things for me, but still microsoft and  
the other biggies will use AJAX, it's there, it's a buzzword, they  
will create Outlook Web App and it will not be safe (at least not  
with todays technology)

Cheers
andre

PS: ... who would not feel safe writting AJAX apps.

On Nov 9, 2005, at 9:38 PM, Dan Shafer wrote:

> I almost labeled this post off-topic since our purpose here is to  
> discuss how to use Revolution. But I decided on balance that it  
> affects everyone here, so I left off the [OT].
>
> I've just posted a blog entry at http://www.eclecticity.com/. 
> 3c66aaec that I believe should be of interest to everyone who is in  
> the programming universe today. I've been leaning in this direction  
> for years, drawn strongly to it for the past few months, and have  
> now tipped over the edge. Some will think I'm over the edge,  
> alright, but perhaps not in the way I intended.
>
> My prediction -- based on a lot of evidence and clinched by two  
> leaked Microsoft memos that you really need to read (they're  
> indirectly linked in my blog entry) -- is that the days of the  
> desktop app are indeed finally numbered. At best, we will see  
> desktops reduced to being containers for ultra-thin clients and  
> specialized Internet browsing tools while *everything else* runs as  
> a (probably ad-supported) Web service.
>
> Yeah, I know. You've heard this before. And there's a lot of  
> skepticism here and elsewhere on the Net. But Ray Ozzie's no idiot  
> and Microsoft's not ignorant or stupid (whatever else they may well  
> be).
>
> Comments welcome, though I'd appreciate it if you'd register for my  
> blog (it's free) and post them there even if you choose to echo  
> them here. This issue is much bigger than Rev but it affects  
> everyone on this list, IMNSHO.
>
> -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-. 
> -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
> Dan Shafer
> Technology Visionary - Technology Assessment - Documentation
> "Looking at technology from every angle"
> http://www.eclecticity.com
>
>
> _______________________________________________
> use-revolution mailing list
> use-revolution at lists.runrev.com
> Please visit this url to subscribe, unsubscribe and manage your  
> subscription preferences:
> http://lists.runrev.com/mailman/listinfo/use-revolution