From: SkyFlash <sky...@ch...> - 2003-01-15 13:21:58
|
>Once you add one, it's trivial to add another. Think of how STL does >this with various _traits classes to get at the underlying types. >You don't need to hard-code things the C++ compiler already "knows" >about. The point is, thats way too complicated. If in two years some new dev reads it and it contains something like that he wont understand a thing. Arianne is not a tech demo. Its not to show we know very complicated highly optimized algorithms. If we use something like that it would have to be a black box, you use it, you have no clue how it works, whatever. But that wont work for something as fundamental as Attributes. Yes it maybe cool for someone who studied computer science to use it but its too much of a hassle to maintain. If we need only floats and strings there is no point in programming arbitrary properties right now. >By what date is it necessary to implement these changes? Just last >week, you told me not to do it. I just need some type of number, not arbitrary values. :) Also I already put them in. :) Works great right now, just I didnt change anything yet to use it. Like the position code for example. >As long as you wrap things up so only the wrapper needs to be changed. >Hopefully "grep -r '\<float\>'" will always return extremely sparse >results :) I actually unwrapped everything in Arianne cause everything was wrapped up in something and no one knew how it worked inside. I dont wanna go to something as complicated if its not needed. What I did was just to place a second map besides the first one. As simple as that. Took one hour and will do what I need right now. Really, I know its tempting to try all that cool stuff you heard about or learned about, but really, someone needs to understand it later. :) Maybe we DO need it complicated in a few month. Could happen. You can still program it anyway and maybe we can put it in a box and seal it somehow inside Arianne so nobody got to fiddle with it. Just, if there are bugs appearing inside, we will be in hell. >I have some time I could use for this, seeing as it's just implementation >code (as opposed to knowing how all the pieces are utilised). I'd rather >see you working on [global] function rather than implementation details >that have narrow, directed, "smells like optimization" scope. I only fix stuff like that if I have to for some reason, in this case I am having problems with the strings cause some were INTs instead of floats and as I gonna change them anyway I can change them to float now instead of changing them to float strings again. :) >What's your position on interior vs. exterior properties on objects? >I.e., is there a use to having access to "all objects with x,y position" >or "all objects with an intelligence attribute"? Not right now... but maybe in the future. >tons of warnings in compiling newarianne, but only one fatal error. >Out of curiousity, why do you not include 'Kyra/' for kyra include files? >(The fatal error was that I have guiExtended/progress.h instead > of gui/progress.h in my .../include/Kyra/. No big deal.) The Kyra dev changed the paths because he moved user programmed code to guiExtended. And as I wrote progress.h thats where it went. :) SkyFlash |