Microsoft *IS* Killing Silverlight
-
CDP1802 wrote:
I would not be sure where to start. The code already has reached some size and explaining every detail may prove to be hard. On the other side, there are still a few open ends, like supporting localization in the XAML markup. I would certainly not mind writing about it, but then it should be something that's useful to the readers.
Heh, I've had the same dilemma. I've never published an article here, but have had several little projects that I thought would be cool to share. The problem was that, barring a ground up rewrite (something I have no time or desire for), the articles would be full of phrases like, "Sorry about this next bit of code, but it does this, I swear," and "Oh like all your code is so pretty." The stuff I do on my own time is usually research type development, and it grows organically as I try new things. I learn a TON this way, but the code is a mess at the end (it's much nicer when I put the knowledge to work on an actual project). Some things I've considered publishing: software 3D rendering (to get the math down, not worrying about acceleration) with some physics simulation thrown in, reverse engineering a DOS game to port to Windows, Stupid DSP tricks... but I really don't have the energy these days. Maybe when the kids move out of the house :) That gives me 14 years, roughly.
David Kentley wrote:
it grows organically as I try new things. I learn a TON this way, but the code is a mess at the end
Yea, that sounds familiar. I started playing around with custom graphical controls stored in a dynamically resizing array in a winapp. Wondering if you've encountered something similar or not. I'm running into some severe non-linear performance issues. A 10X10 array renders almost instantly. A 25X25 array is about 6 times bigger and takes about 6 seconds to render. A 100X100 array is 16 times bigger, so if it was linear, should take less than 2 minutes to render. Over 3/4ths of an hour later, it starts rendering and unlocks the program just short of an hour later. (Not using threading logic.) This is on a dual core, 2GH device running at about 56% CPU while it is calculating locations and rendering images. So, one of the CPU's is running at 100% capacity for about an hour. That's all because of the custom images and event handling being added to the controls as they are being created and rendered. Rather frustrating when the underlying app can recursively hit 33K nodes multiple times in 62 milliseconds.