Call me old fashioned, or possibly slow, but for some reason I never seem to be using the latest version of a piece of software. Until recently I was doing all my .NET work with Visual Studio 2008; this was because VS2010, bless it’s heart, felt sluggish to me.
With the pending release of Visual Studio 2012, which as I write this is available for a free download as a Release Candidate, I decided I’d bite the bullet and start switching. This was also because I wanted to dip into XNA, and As near as I could tell the latest version only worked in conjunction with VS2010. I had to reinstall Resharper to get proper VS2010 support, since I had installed Resharper before I installed VS2010, and after applying my own preferences to both Visual Studio as well as Resharper, I was able to get back into coding. (Am I the only person who hates the preferences IDE’s have to automatically complete parentheses and braces and stuff? I always find myself typing the ending parenthesis, ending up with double, so I delete the extra ones, then I forget where I was in the nesting; and if you get used to that behaviour, suddenly you find yourself not typing ending parentheses in plain-text editors. You can’t win! I’m not a big fan of that sort of autocomplete; Actually, I don’t really like any form of autocomplete, but that’s sounds like material for another post altogether.
The End result is BCDodgerX, which is available on my main downloads page. It is essentially a rewrite of BCDodger, with an unimaginative X added onto the end that means pretty much nothing.
Overall, VS2010 is actually quite good. Call it a belated review; I almost purposely fall several versions behind for some reason. I cannot say I’m overly fond of the use of 3-D Acceleration within a desktop application, but at the same time all the Controls still have the Windows Look and Feel (which is my main beef with Java’s Swing libraries, which have a look and feel all their own), and the desktop itself is accelerated with Aero anyway so I suppose it’s only a natural progression. (Besides, I don’t play games very often and this 9800GT should get some use…).
The tricky question now is when I should start migrating my VS2008 projects to 2010, and whether I should switch to the latest framework. I can switch to VS2010 without using the latest framework, of course, but I wonder what benefits I will see? One day I’m sure I’ll just say “screw it” and open say, BASeBlock in VS2010 and jump in; I’m falling behind, after all (What with the aforementioned release of 2012 on the horizon). And VS2010 is definitely an improvement both tool and functionality wise over 2008, so there isn’t really a good reason not to switch now. No doubt I’ll keep making excuses for myself. Oh well.
At first, I thought I hated XNA; but now I know that what I actually hate is 3D programming. I imagine this is mostly because I got extremely rusty at it; additionally, I had never truly done 3-D programming, at least in the context of a game. My experience at that point was pretty much limited to the addition of 3-D graphic capabilities to a graphing application that I wrote (and never posted on my site because it hasn’t worked in ages, is old, and uses libraries/classes/modules I have updated that are no longer source compatible etc.). Of course that didn’t have constantly changing meshes, used DirectX7, and it was shortly after I had finished that feature that I abandoned the project, for whatever reason. I had never dealt with 3-D in a gaming capacity.
The purpose of XNA is to attempt to simplify the task of creating games- both 3D and 2D, for Windows as well as XBox 360. And it definitely does this; however you can really only simplify it so much, particularly when dealing with 3D Programming. My first quick XNA program was basically just to create a bunch of cubes stacked on one another. This is a very common theme given the popularity of games like Minecraft, but my goal was to eventually create a sorta 3-D version of Breakout (or, rather, BASeBlock 3D).
I was able to get the blocks visible, after a lot of cajoling, and doing the work on paper (Visualizing 3-D space and coordinates are not my forte). But it ran at 10fps! This was because I was adding every single block’s vertices to the VertexBuffer; for a set of blocks in a “standard” arrangement of, around 1920 blocks (which is probably a number that would make the 2-D version go around 10fps, to be absolutely fair here), that is over 11520 faces, each of which actually consist of a triangle list of 6 points (I tried a triangle fan but it didn’t seem to even exist (?), oh well) meaning that I was loading the VertexBuffer with over 69120 texture-mapped vertices. That’s a lot to process. The big issue here is Hidden Surface Removal; obviously, if we had a cube of blocks like this, we don’t need to add the vertices of blocks that aren’t visible. I’ll admit this is the part I sort of gave up on that project for the time being; that would involve quite a bit of matrix math to determine what faces were visible on each block, which ones needed to be added, etc based on the camera position, and I kind of like to understand what I’m doing, and I, quite honestly, don’t have a good grasp over how exactly Matrices are used in 3-D math, or dot products (at least in 3-D), and I prefer not to fly blind. So I’ve been reading a few 3-D programming books that cover all the basics; the book itself I believe goes through the creation of a full 3-D rasterization engine and has a lot of in-depth on the mathematics required; this, paired with concepts from Michael Abrash’s “Graphics Programming Black Book” should give me the tools to properly determine which blocks and faces should be added or omitted.
Anyway, scrapping that project for the time being, I decided to make something 2-D; but since I was more or less trying to learn some of the XNA basics, I didn’t want too much of the concepts of the game itself getting in the way, so I chose something simple- I just re-implemented BCDodger. I added features, and it runs much better this way, but the core concept is the same.
Some minor peeves
XNA is quite powerful- I have no doubt about that. Most of my issues with it are minor. One example is that XACT doesn’t seem to support anything other than WAV files, which is a bit of a pain; this is why BCDodgerX’s installer is over twice the size of BASeBlock’s, despite having far less content). Another minor peeve is that there is no real way to draw lines, or geometry; everything has to be a sprite. you can fake lines by stretching a 1×1 pixel as needed, but that just feels hacky to me. On the other hand, it’s probably pretty easy to wrap some of that up into a class or set of classes to handle “vector” drawing, so it’s probably just me being used to GDI+’s lush set of 2-D graphics capabilities. Another big problem I had was with keyboard input- that is, getting text entry “properly” without constant repeats and so forth. Normally, you would check if a key was down in Update(), and act accordingly. This didn’t work for text input for whatever reason, and when it did it was constrained to certain characters. I ended up overriding the Window Procedure and handling the Key events myself to get at Key input data as needed, then hooked those to actual events and managed the input that way.
Overall, I have to conclude that XNA is actually quite good. There are some intrinsic drawbacks- for example it isn’t cross platform (to Linux or OSX), and the aforementioned basic problems I had, which were probably just me adjusting my mindset. It’s obviously easier than using Managed DirectX yourself, or DirectX directly (if you’ll pardon the alliteration), and it is designed for Rapid creation of Games. With the exception of the High-Score entry (Which took a bit for me to get implemented properly) BCDodgerX was a single evening of work.