Vertex meshmash

Spent several days trying to get this

into this

And let me tell you: it was no walk in the park. I spent days on this shit. Until I found the answer. On stackoverflow, naturally. As always. The fix took one minute and eliminated 200 lines of horribly ugly experimentation code that got me nowhere. Urf.

On the upside I got some good debugging tools in place, and extended my Maya import pipeline. Now I only need to finish a couple of levels, icons, splash screens, vehicles and sounds. Mhm.

This whole game was supposed to take a couple of days to make, in fact that was the whole point if it. This is definitely a case of better never than late, as the game will be too simplistic to attract anyone to pay $1 for anything but to uninstall it. (New business model right there, BAM! I just keep on delivering.)

Must... control... language study

HAI I HAS A ANIMAL GEMMEH ANIMAL BOTH SAEM ANIMAL AN "CAT", O RLY? YA RLY VISIBLE "J00 HAV A CAT" MEBBE BOTH SAEM ANIMAL AN "MAUS" VISIBLE "NOM NOM NOM. I EATED IT." NO WAI VISIBLE "J00 SUX" OIC KTHXBYE


This has gotta stop, I've been looking into LISP vs. Python too long. Now I'm off to Maya. (Thanks to fullvolume.co.uk for LOLCODE interpreter.)

Study

In between contractor work I've done a couple of days worth of study. Programming that is. I usually do between jobs, and since I've had a few in my career (about 10, depending on how you count) it's been a few dives here and there. Usually on something with buzz, sometimes cutting edge stuff, occasionally some bleeding edge tinkering.

I'm one of the late early adopters when it comes to software in general and programming tools in particular. I did C a couple of years after it became a standard and C++ not very long before it became an ISO standard; Linux from pretty early on; Python as it was becoming fast, simple and useful; git as it overtook the other free, distributed SVCs; iOS as companies started to take an interest.

The advantage of being late is that you see where things are heading. But over the last years three software fields have been expanding massively, but I haven't hardly paid attention. They are HTML, Android and Java. All of these are technically flawed one way or another.

• HTML is separated into structure, style and logic. The structure is that of hierarchical boxes-in-boxes, pretty much like any layout manager since the 80's. Sometimes you use CSS to move a box out of it's parent and place it relative to screen or page, but generally it's boxes-in-boxes. If you want to make changes to a layout you usually have to spend horrid amounts of time to move a box into a different parent. Both for layout/style reasons, but also for logic reasons, and especially so for single page applications as load order comes into play with Ajax, etc. And browser/IE compatibility.
  
  Performance-wise it's really bad. The traditional static web page is not very responsive: when you click a button, the next page can take seconds to appear. Seconds! In the year 2014. Usually there is also some flickering involved when loading a page. But single page applications is usually not much better. Response times when clicking is usually also in the seconds, and on almost all sites you see some icons moving a couple of times during load, and text is pushed down while other components load, etc.
  
  Technically, HTML+CSS+JavaScript is too "open", you can solve a problem in any number of (bad) ways, which leads to a plethora of frameworks to assist you. And as companies struggle to use something which will survive a decade, they grasp for JSP, Java Servlets, ASP .NET, Rails, Python Django, Node.js, Django, GWT, Backbone.js, CoffeScript, Angular, jQuery and a ton of others. This alone should get everyone thinking "what's wrong with this picture", but strangely enough status quo is all I hear. (My theory is that most of the good people managed to stay out of the HTML slough.) Frameworks in general are evil things, but on a quagmire tech platform perhaps it's the least bad option.
  
  But worst of all, worse than all of the technical awfulness, is that HTML design allows you to make all pages unique. So after 37 signals published their series of articles on A/B testing, small software entrepreneurs en masse set out to make long pages with round photos where content is split up into easy-to-read chunks, starting with a link to an easy-to-read pricing plan and ending in a footer where you can reach the other products. Good for them - uniformity is a good thing. Both for your eyes and your wallet. Apple understood this and perfected a one-size-fits-all layout-and-logic engine in iOS. This is what the customer wants, and this is what the suppliers should deliver. Unfortunately, most site's still look like something the cat dragged in, but in a unique way. The overall lack consistency between sites is why pretty much all HTML pages look like shit, no matter how well-designed they are. That's why it's important to have a type of iOS-ey template to stick to.
   
• Android. Well what can I say. It's still miles behind iOS in almost all ways possible, but it's open and gratis. In the long run I suppose Android could win, but right now it's too much compatibility hassle, too much design uniqueness (same flaw as HTML), and way too much code+work to make an app that looks great on all/most/any target device.
   
• Java. Verbose, slow and more often than not comes with huge do-it-all frameworks. Which are great for hellowdy world, but not so good when scaling up. Sure, Java takes the mundane and error-prone out of C++, but it also adds a virtual machine, byte code and usually a build system or two including dependency management. And the everything-is-an-object just gets on your nerves as much as the verbosity after some time. Steve Yegge nailed it with "you can't trust anyone who loves it too much." Oh so fair, candid and right the guru is.

Now it's gotten so far though that it's even hard to find contractor work unless you're prepared to get your hands dirty with the above technologies. So I started looking into React JS for the client-side part of single page webapps. It sort of takes away some of the ugly disconnects between HTML and JavaScript, and it does it fairly well I think. You can read about my conclusions here. It still sucks ass though as most of the above still applies.

Interestingly, with React JS I would be a fairly early adopter if I started using it now. Especially as the community around Angular JS is dwarfing the React community, no matter how bad Angular is in the long run. But I'm not going to. It's always sad to see the better (or least bad) party go down, but it wouldn't be the first time.

Changing the status quo

What I would like to see in the future to replace the current state of affairs is some new type of full-screen OS desktops replacing the windowed dittos in Windows/OSX/X but loaded from a server backend. Design should be fairly strict by default, it would take some effort for custom designs (like iOS), not the other way around (like HTML). Communication should be REST/JSON, so you don't have to know more than the four HTTP methods as RPC is a bitch as 100+ operators always gets messy, requiring murder. Documentation is murder. The language doesn't matter as long as it's run native without a virtual machine bogging things down, and it must have good tooling. Which excludes a lot of junk. Portability would have to be built-in; and client and server code would be similar or same.

If this type of system was easy to install and also contained a great standard library (like for instance Python) it could potentially kill HTML, Android and iOS development along with .NET/Java VMs in a single blow! And what a great stack to start out with: the feeling of native without the hassle; the intrinsic performance of native; a standardized and beautiful fullscreen design; and the distribution and connectivity of the web.

It needs to contain a compatibility layer for windowed apps, but with that in place it would easily replace Explorer.exe, Dock.app, KDE, Gnome or whathaveyou. If it contains the basic web surfing (i.e. Chrome in fullscreen) and it trajects above the viral threshold, this type of tech could eliminate the ugliness of HTTP once and for all. We'd all have design restrictions, high performance and good tooling. Thus creating good-looking, distributed, fast applications would be fun again (and the end-users would love it)!

This type of system will be to software what IP is to communication. (To bad we're all still running RS-485. Sure, sure it's better than what came before! Yes, yes!)

Please Linus Thorwalds or Mr. Equally Magnificent Programming Guru, build this über-system now!

After the fact

Time to recap my last contractor work. The last project was my third contractor assignment.

• On the first assignment I did good work, swift and professional. My hands weren't tied and I didn't have a boss.
• On the second assignment, I got sick of the bureaucracy in the organization; the highly unprofessional wannabe architecture astronaut; and the frameworks in the tooling.
• On this third last assignment I feel that most were professionals with good understanding of what needed to be done; but they were all pretty stiff. The management on the other hand didn't have a clue on where to go in the future, which is reflected pretty much by their upcoming downfall (remember where you read it first).

I realized that I prefer stiff professionals above incompetent but lively co-workers. I had a pretty good time on this last day-job, even though it was my first maintenance work. It was almost only maintenance as we had no mandate to improve the product! I could only look at this "strategy" with disbelief, mouth gaping. Have you ever heard anything that dumb? The management uses the antithesis of "Get Out of Their Way". Need I say that the employee turnover is humongous?

I knew this before, but Java remains an eyesore for me; and a camel i still a horse designed by a committee. Java need a runtime. Java is verbose. Java is slow. Java comes with environment variables; something called classpath (which resembles system path, but not really); some ugly home-brew .properties files for static settings; a bit of .xml files when you're building some server; a couple of frameworks; a slow and buggy IDE (often Eclipse which is designed by a committee, using plugins built by amateurs); a slow verbose build system, often using plugins from amateurs. Usually, to spice it up, there's some legacy parts that needs to be involved, for instance Java Swing for building GUIs or GWT or JSP for building web.

This is of curse disturbing, but more disturbing is that most co-workers I met when working on Java project don't even know this. In fact, I've not met plenty of seasoned professionals whom have never worked with any language other than Java (perhaps apart from the scant Haskell class in school). If you've never lived in a house, it wouldn't seem preferable to living in a cave. The basic things are there already: walls, roof, security, windbreak.

I'm not saying that it's an option for building enterprise-scale web servers, but using Python you have all of the things you need built right in. 10 lines of code, no extra configuration and you have your minimal web server.

But if you've never seen anything else than hundreds of lines of code and configuration; installation, paths and frameworks; to actually get a web server to square one, well then you're still inside a cave. And you can read all the fact books about caves you like, but you're never going to get out unless you try living in a better place for a while.

I have no new assignment yet, but I actually hope I get to do a bit of web front-end next time. Not because I think it's any good (spoiler: I don't!); I just need some variation. But more than this, I want a job without a boss. It's fantastic and natural to be creative and not held back by strategy/stupidity.

Now it's off to the dentist, then I'll get my hands dirty with level-making for a couple of days.

Explosions

I've been a bit lazy lately with this blog, but here's the basic idea of the next game anyhoo:

Over the last couple of weeks I've spent horrible amount of hours on trying to get the projection of UVs for the texture right, so that I could have shadow volumes on the ground. It was a lot more complicated than I first thought. Eventually I settled with a fixed width/height ratio (4/3 for iPad), I wasn't able to pull it off with a 4/3 texture in any sensible way. As you see I couldn't get the uvst algorithm right, neither with quads nor with triangles (I suspect OpenGL uses the same algorithm for both). And it was pretty difficult to model the ground in Maya. Must be crazy doing all this work for a bunch of horribly CPU-intensive shadows. Anyhoo. Now it kinda works and looks like shit. Remains to create some levels I guess. And launch.

I'll probably go with the (infinitely stupid) free+in-app purchase model, mostly because the game won't be good enough for anything else. Next game I'll make a real good one, seven being my lucky number n all.

Done and done

Bound 3D was reviewed and passed last night. It has yet to receive interstitial iAds, but I looked pretty long and hard for it and eventually I found a small clause inside the iAd App Network FAQ (gosh):

Your submitted app will be reviewed by iAd to determine the app’s appropriateness for receiving ads from iAd advertisers. After your app has been approved, you can start earning revenue for ads that run in your app.

Apart from that the app seems to work. I feel pretty proud about the app, but don't know why exactly. I think it may be because for the first time I made an app that fits the format. It looks peculiar, I'd say it's appalling in that mid-90'ies 3D simulator kind of way. But the feeling is right. Oh, and it's boring. Alas, one out of three is my best achievement so far so that's probably why I feel a bit proud of myself now that it's in the open. Over to the post mortem.

What went wrong		What went right
Booooring There's no fun to be had there. You get no gratification when you complete a level, you just get an extra ball. It's not very relaxing, in fact you get a bit stressed by the whole "hurry before the balls get over here" thing. On the other hand, there's no thrill either. The worst thing that can happen is that you get your line shattered. It's all very stiff, so there's no amusement to be had. This is not a game, it's a recipe for apathy.		Feeling Swiping works as I wish it would. The symbolics could be better. For instance, I just realized that everybody else in the iOS world uses dashed lines for cutting. I use a solid line. Apart from that I'm satisfied with the overall feel.
Look It looks like a home-made game from the mid-90'ies. I wrote 4k intros in the mid-90'ies that looked better than this. It's not cute. It's not sharp. It just looks like... glahh...		Small 2725 lines of code, 12 lines of configuration (for the ball), one mesh+one texture (the soccer ball itself), 16 calendar days. Goes without saying that it was my best game so far in terms of development hours. In terms of dollars per hour it still sucks ass.
"Buy full" A full what? A full moon? A full stomach? Full som en kastrull? And if you read between the lines, there's absolutely no explanation of what you're buying if you would attempt to buy the full version or what it would cost, until after you've entered your password. Who would ever do that? (PS. with "buy full" you're getting rid of ads and getting more level shapes than the default cuboid.)		Ads It was interesting to see Apple's take on ads. The interstitial ad, that I used, is a full-screen ad that only shows on the iPad. With some games I enjoy having a break with the full-screen ad. Especially in dull games.
Rotation Half of the time, the level is rotating in the opposite way that you'd like it to move. And there is only the indirect way of rotating a level by cutting off a piece from the opposite side of the level. Who comes up with these things, huh?		Goal Somehow I feel I have explained the goal of the game well enough: reduce the level's volume to 15% or less to finish the level.
Explanations I explained the goal of the game well enough. But the rest of the information printed on screen is hardly: noticed, visible, interesting, nor relevant.		Riding on giants Hm. I meant midgets. I was inspired by the game Trap on Android and the game Addictive Balls on iOS. None of them are any good, and I wasn't even able to match any of them.

What next then? Maybe I'll celebrate. Eat a biscuit, for instance. I can soon buy those with all the money I'm makin'. Yeah - bitch!

Next game

Was thinking of revisiting my old new idea, but after a few hours of tweaking I understood why it must be left alone.

1. iOS is my only current platform (the only one where people pay for stuff),
2. Cyber Sled was all about the dual joystick controls,
3. touch emulated joysticks are a nuisance.

So my next game which I'm interested in doing, and which I feel could become fun to play, is a point-and-tap 1st person shooter. It doesn't require any further engine development that I'm aware of, and it sort of resembles Kill Cutie in a way. But not too much.

The basic idea is it's a single-player game where you're up on an outpost with a rocket launcher. You're the "city colonel" or something of the kind, ready to defend your average American town against the communist invasion. The backdrop might be made up from photos... The cars are all beige, except as we all know, the commies' only roll in red cars.

Your goal is to tap the indicator. I know, I know, I'm on a slippery slope. My previous game was all about swiping. Now it's about tapping. If this continues, my next game will by about staring. But it won't.

This time, I'm going to set some goals. What is going to be fun with this game?

• Super simple controls,
• a'shootin',
• explosions,
• sending foreigners to the American dream Hell,
• when you get tired of playing you go collateral ftw,
• watching the little piss-ants panic when there is a road block up ahead and someone has a bigger gun than them.

As I'm thinking about it, I assume I need the texture mapping should use some scaling matrix on the XZ-coordinates. Apart from that, the technical stuff should be very easy to pull off now that I have the resource loading, some cars, the explosions, the spawn and remove points and an ugly menu system. This game will be relatively easy to do (apart from all the backgrounds, cars, icons, splash screens, music and screen shots), but I need to get some games out there.

After the initial 57 sold copies of Downwash in November, I sold 3 in December. So it will be free, and I'll have a single in-app purchase for the additional X levels, unlimited ammo and unlimited fire rate. Or maybe I'll do a simple Tekken clone, who knows.