View Single Post
Old 09-30-2006, 07:21 AM   #10
ali
Addict
ali doesn't litterali doesn't litter
 
ali's Avatar
 
Posts: 302
Karma: 116
Join Date: May 2006
Device: Iliad, dude!
Quote:
Originally Posted by rlauzon
Yes, they will. When they announce that they device is no longer in beta test and is "official". I expect that to happen by December.
I can't imagine that to happen. Think developer guy goes to marketing guy and management guy, and he says "That's it, we're done". Now there's two options: First, publish it, giving users a reason to sue or get refunds; where several already said they'll loose their patience in reaction to that announcement. Option 2: Say nothing, stay in the state of no liability because users signed they can live with unfinished software.



Quote:
Originally Posted by rlauzon
Yes, they will. When they announce that they device is no longer in beta test and is "official". I expect that to happen by December.
You are confusing what they are offering for the 2.7 release with what they are going to offer after that.[/quote]
Nope. I just know something about how these things work, so I made a guess. When 2.5 came out, I predicted (and wrote it in these forums several times) that there will be some more improvement, and then just mini-steps, with a long-term result of 10 hours. So far, it happened precisely how I figured it would.

Quote:
Originally Posted by rlauzon
And that old lesson became history when more modular software designs were used.
Woo. Not at all. The benefit of modular software design is that you can get lots of features with small development effort, in short time. It speeds up everything because you can use stable, tested, general-purpose code without having to worry for details.

The downside (and this was also tought in my software engineering courses, btw) is that the resulting code is suboptimal, and breaks down when your system has unusual constraints that were not known when the modules were developed. You get stuck in the situation where improvement is only possible by refactoring all of the general-purpose modules for the new constraint. (Say, skim through xpdf, gtk and matchbox to see where floating-point arithmetic is used out of laziness) None of the Iliad's modules was written for a system with a .5 Hz screen refresh rate, and hence, there is complex, frequently failing code to handle the screen refreshes. There is a full-blown window manager running that can draw windows, window borders, move and resize windows, maintain stacks of partially visible windows - but all of these features are useless, though they waste energy. (If you are good at working the stylus: A double-click-and-move with the stylus onto the invisible border of the toolbar window will move that window.)

That's the problem: All of these things just waste a little bit of energy and run-time, but it sums up. And the only way to get real improvements is to rewrite them all.
ali is offline   Reply With Quote