Richard and Andrei: do you remember (if you heard of it) the book called "All the way with GTIA?" My parents bought me a copy of it, but I never really used it as I bought an ST a month or so later.

The GTIA added several extended gfx modes (including some 16 colour ones at 80 x 224). It was only available on the XL and XE series of machines IIRC.

I don't think the GTIA was responsible for DLI (Display List Interupts), I think they were in the earlier CTIA chip too.

I started with QBasic and bombed royally; I could make simple text games and still graphics but animated sprites were impossible for me. Took up C++ with an interest in making it a career, ended my education not feeling much smarter. We used Visual C++ and played with MFC a lot, so I went looking for something a bit more "bare" C++ which ended up being DJGPP. The download page mentioned Allegro, I thought "cool" and downloaded that too. Rest is history. Since then I've also picked up Win32, OpenGL, some Java .. Allegro is simple video game easy mode though.

I used Recreational Softwre Design's Game Maker (don't think this has anything to do with the current one) for a while and made several platform and overhead games ~1995. Meanwhile I was learning C and programming text-based games (compiled using Turbo C++ 3.0). I had a book called VGA Game Programming Explorer which came with a disk with C code on it to handle various things related to programming VGA games.

I didn't know at the time that the code from the book would compile with my Turbo C++ compiler because I thought TC++ would only compile C++ code. When studying C I learned that C was compatible with C++ and figured out how to compile some of the code that came with the book. I started hacking around with the example programs and learning how to make the code do what I wanted.

My first C games used VGA Game Programming Explorer code for the graphics and MikMod for the music/sound engine. My first year of college I decided to give up all my pirated software so I deleted my TC++ compiler and wouldn't do any more games except experimental stuff with Ada which I was learning in one of my classes. The graphics library sucked for Ada so I went looking for a better one to use with Ada and stumbled upon Allegro and DJGPP. DJGPP intrigued me as it was a complete compiler for free, 32-bit, gave me access to a lot more RAM than TC++ did.

So I downloaded DJGPP and Allegro and started immediately making use of the new tools. Allegro was so easy that I was up and running in no time. I converted one of my old Turbo C games to use Allegro in an hour or two and Light Breaker 32 was born. I began work on a tilemapping engine almost right away but I wouldn't make anything with that for another year or two. I was actually working on a suite of game-making tools in the vein of the old Game Maker that I had used before so both me and my older brother could make some games together. These never came to completion. Later, I ended up using the tilemapping engine to make Paintball Party.

When I first started using DJGPP it was MikMod still for music and when Allegro became cross-platform I used FMOD in Windows and MikMod in DOS. I dropped DOS support partway through the development of Paintball Party so I used FMOD exclusively for a while. When DUMB came out I switched to that and have stuck with it ever since. I also added support for MP3 and OGG music to Paintball Party after a while and used AlMP3 and AlOgg respectively for those.

When OpenLayer came out I started playing around with that. I made a more advanced tilemapping engine taking advantage of hardware accelerated features. I had one game in the works using OpenLayer but I kept running into problems with it. When 2.0 came out I was excited but the excitement quickly faded as 2.0 was more broken than the first one. Many of the things that were broken have been fixed but I still can't get it to compile properly with the options I want.

To sum up:

• 1995: Game Maker (Recreational Software Design)

• 1995: Turbo C++ with VGA Game Programming Explorer and MikMod

• 1998: Ada and it's crappy graphics lib (never made any games with it)

• 1999: DJGPP with Allegro and MikMod

• 2000: MinGW with Allegro and FMOD

• 2002: DUMB and AlMP3 and AlOGG

• 2006: OpenLayer

• 2006: Back to plain Allegro + DUMB + AlMP3 and AlOgg where needed

I disagree. I think it lost its magic for a long time, and then they invented pixel shaders! Sure the magic is more about maths now than engineering, but it's still magical.

I think the magic has now shifted more towards the software side of things. But i agree that pixel-shaders are cool, but the magic isn't so easy to spot as the effects you could do on the old 8-bit hardware.

AE.

Restrictions can make you have a narrower (but clearer) game design or idea. And then there is the satisfaction of getting around a technical hurdle. Perhaps you just had to have been programming on those machines to appreciate it..

And pray tell, what makes it impossible for you to exceed the current limits in the same way? All I hear is curmudgeons complaining that it's too hard to stretch the limits of modern hardware, and/or being sour about it being too easy to do now what they took considerable pains to do 10 years ago.

Well, it's not impossible, just mostly futile on the PCs.

In the earlier days of computing a given platform stayed on the market in an almost unchanged form for years at a time. This gave developers time to improve and research new, clever techniques. And this work actually had some use, because they could then market the product using those improved techniques.

These days, you can see these taking effect on the consoles, mostly. If you look at launch tiles and "last generation" titles for any given console, you will invariably find that the latter have much better graphics, despite running on the same hardware - so the magic is still there.

On the PCs however, pushing the limits of a specific line of GPUs has little sense, because by the time you perfect your optimizations, that line is already obsolete. This doesn't pay off in the long term, so it's usually not done.

For the hobby developer, there are also other concerns:

• The old-timer magical gfx tricks often used very specific quirks in a given piece of hardware. This is nigh impossible on PC, because you'd have to write custom code for every permutation of computer parts.

• Also, these days, the bar is set much, much higher if you want to push the limits. The knowledge needed to devise a cool trick on the C64 or Amiga was not that vast. If you want to do something really great now, you have to have a very good grasp of mathematics, and the hardware itself is much more complicated. Therefore, the "magic" is less accessible to a hobby programmer.

I'm not saying that programming lost its magic on new hardware. But it's definitely a different type of magic now, and not all of that difference (but surely some of it) stems from nostalgia.

You set a graphics standard for your game once and then don't care about progress.
As a small developer you shouldn't try to compete on the eye candy side.
Focus on the things that make games popular for decades instead, gameplay.

It's nostalgia for the experiences of the joy of getting it all to work. Also, even though the hardware has long since become obsolete, the experience I gained from programming for it was the foundation for experience that I gained later on.

AE.