Comments on: To CLOS or not to CLOS http://nklein.com/2009/06/to-clos-or-not-to-clos/ software development and consulting Thu, 15 Jul 2010 11:51:39 +0000 http://wordpress.org/?v=2.9.2 hourly 1 By: pat http://nklein.com/2009/06/to-clos-or-not-to-clos/comment-page-1/#comment-171 pat Mon, 20 Jul 2009 14:48:46 +0000 http://nklein.com/?p=624#comment-171 There are probably ways. I would have to experiment quite a bit to get a way that I felt confident would work throughout. I have since been reconsidering my stance on this. I think I'm still going to use classes and generic functions for things like cameras and meshes that I expect to get called only a couple of times per frame and use structs for things like facets and vertexes that will be referenced hundreds of thousands of times per frame. I still haven't fully embraced Agile programming, but I have done enough fundamental changes to APIs in my time, that I'm reasonably confident I can go back and retro-fit whatever I need to if some portion is weighting me down. There are probably ways. I would have to experiment quite a bit to get a way that I felt confident would work throughout. I have since been reconsidering my stance on this. I think I’m still going to use classes and generic functions for things like cameras and meshes that I expect to get called only a couple of times per frame and use structs for things like facets and vertexes that will be referenced hundreds of thousands of times per frame.

I still haven’t fully embraced Agile programming, but I have done enough fundamental changes to APIs in my time, that I’m reasonably confident I can go back and retro-fit whatever I need to if some portion is weighting me down.

]]> By: Grant Rettke http://nklein.com/2009/06/to-clos-or-not-to-clos/comment-page-1/#comment-169 Grant Rettke Mon, 20 Jul 2009 06:22:34 +0000 http://nklein.com/?p=624#comment-169 Isn't there a way you can write your Lisp code so that even if you start out using structures you can make it very easy to migrate to objects later on? Isn’t there a way you can write your Lisp code so that even if you start out using structures you can make it very easy to migrate to objects later on?

]]> By: pat http://nklein.com/2009/06/to-clos-or-not-to-clos/comment-page-1/#comment-132 pat Tue, 30 Jun 2009 01:12:59 +0000 http://nklein.com/?p=624#comment-132 The structs are five times faster. Certainly, both are fast enough for almost everything. But, for realtime, interactive 3D stuff, there is enough going on that some of those 5x's could make a big difference. Certainly, I don't want it to only be able to do 1/2 the resolution or a quarter of the polygons that the C++ version does if I can help it. Yes, this falls in the "premature optimization" category. But, it also falls in the "easier to do now" and "just about as pretty" and "some of it will have thousands of accesses per frame" categories. The structs are five times faster. Certainly, both are fast enough for almost everything. But, for realtime, interactive 3D stuff, there is enough going on that some of those 5x’s could make a big difference. Certainly, I don’t want it to only be able to do 1/2 the resolution or a quarter of the polygons that the C++ version does if I can help it.

Yes, this falls in the “premature optimization” category. But, it also falls in the “easier to do now” and “just about as pretty” and “some of it will have thousands of accesses per frame” categories.

]]> By: Zachary Beane http://nklein.com/2009/06/to-clos-or-not-to-clos/comment-page-1/#comment-131 Zachary Beane Tue, 30 Jun 2009 00:56:32 +0000 http://nklein.com/?p=624#comment-131 So they're both so fast it doesn't matter which you use, but you're going to go with structs? So they’re both so fast it doesn’t matter which you use, but you’re going to go with structs?

]]>