Anyways – the industry is not quite as bad. I did see reuse happen, once or twice

Kidding aside, our current game builds completely on the foundation laid by the previous one. Doesn’t mean we’re not ripping out large parts, and replace others. We also do use internal libs for common things. So you might say we’re slowly getting into a sane place. (It happens at other companies, too. The rising costs force people to reuse)

Groby’s second point is something that I think is really tragic (not with respect specifically to SDL — it’s clear that there are real reasons why one might not use that.)

I was friends with a developer at a huge software house working on a real time strategy game…ok, it was Activision, and the game was Dark Reign. And he described to me the game, and that he was working on the AI. And I said “Wow, that’s great that it’s an RTS, because presumably you can basically use the same harness as all the other 3,000 RTS games and just slap on new art and stuff and rework the AI, right?”

It was then that he explained to me that no one in the game industry ever reuses any code, ever. No matter how good the code from the previous project is, the very first thing the 23 year old hotshot you hire to work on the next project does is say “this code sucks. I’m going to rewrite the [render engine / parser / scripting language] from scratch. Then it will be better!”

That was the day I decided I was happier playing games than developing them (for money, anyway). Now I develop distributed filesystems. We reuse code.(*)

Maybe I exaggerate.

But not much.

(*) please don’t ask me about the one specific place where we’re egregiously not reusing code.

#1 – Legal departments have problems with OpenSource licenses. Sure, the official explanation is “Who do we sue if it contains IP that doesn’t belong there”, but that applies to commercial stuff too – most EULAs just deny responsibility. Then again, I’m not a lawyer, nor do I play one on TV.

(Sidenote – ActiveState is making a lot of money out of repackaging OpenSource stuff (Perl, Python) and adding support and a liablility waiver on top of that – there’s a business opportunity, if anybody wants to go for it.)

#2 Hey – we’re the industry that invented “Not Invented Here”. There’s plenty of guys out there who would refuse to touch any code that hasn’t been written by the game team itself.

#3 Cross platform is expensive. SDL will not take care of all the issues in development. But even worse, development is only part of the cost. You need to do compatibility testing for the other platforms, write installers, manuals, etc.

#1 & #2 can be fixed quite easily, but #3 is a problem.

Of course, SDL is a wrapper rather than a platform, so it is always destined to be (at least) one step behind DirectX on Windows, which doesn’t help matters.