3 ms·

Ooh, this is neat. Long time since I've had to read assembler. The first thing that jumps to mind is wanting to read the tests for this, which of course don't e

Ooh, this is neat. Long time since I've had to read assembler. The first thing that jumps to mind is wanting to read the tests for this, which of course don't exist. In retrospect, creating this seems herculean; there's so much knowledge about how it should work that lives only in somebody's head, and only as long as they can hold it all there.

jonny_eh 5y ago
The nice thing about game development (at least back then) is that once you ship the game, you can expunge all that info from your brain. So as long as the game works when you ship it, no need for tests to catch regressions.
- degenerate 5y ago
  That is, until your boss says "Great job everyone! Time to get started on Pinball Fantasies 2"
  - jonny_eh 5y ago
    Better hope it fails :D
- pavlov 5y ago
  This model of development also meant cross-platform ports were practically rewrites. In practice it wasn't really possible to reuse code unless the target platform had the same CPU and at least equal graphics features (as was the case with e.g. porting from Atari ST to Amiga). So every port started from scratch.
inDigiNeous 5y ago
Well in a way it was also much more simpler, you had in DOS days a direct interface to the video memory and you most probably understood more accurately how the whole operation worked. So developers could map the hardware in their head more easily, vs nowadays where you have in many cases at multiple layers of abstraction between the code and the computer, having to figure out cryptic black boxes like OpenGL drivers and what not that are difficult to debug and make it hard to understand how things work. Back in the day of course challenge was more not having widespread documentation and figuring things out as you went, especially with these pioneers.
- bzzzt 5y ago
  Instead of documented OpenGL drivers which could have cryptic bugs they used undocumented VGA modes which could have cryptic bugs. Seems nothing's changed then ;)
  - inDigiNeous 5y ago
    Well thats true :)

jonny_eh 5y ago

The nice thing about game development (at least back then) is that once you ship the game, you can expunge all that info from your brain. So as long as the game works when you ship it, no need for tests to catch regressions.

degenerate 5y ago
That is, until your boss says "Great job everyone! Time to get started on Pinball Fantasies 2"
- jonny_eh 5y ago
  Better hope it fails :D
pavlov 5y ago
This model of development also meant cross-platform ports were practically rewrites. In practice it wasn't really possible to reuse code unless the target platform had the same CPU and at least equal graphics features (as was the case with e.g. porting from Atari ST to Amiga). So every port started from scratch.

inDigiNeous 5y ago

Well in a way it was also much more simpler, you had in DOS days a direct interface to the video memory and you most probably understood more accurately how the whole operation worked. So developers could map the hardware in their head more easily, vs nowadays where you have in many cases at multiple layers of abstraction between the code and the computer, having to figure out cryptic black boxes like OpenGL drivers and what not that are difficult to debug and make it hard to understand how things work. Back in the day of course challenge was more not having widespread documentation and figuring things out as you went, especially with these pioneers.

bzzzt 5y ago
Instead of documented OpenGL drivers which could have cryptic bugs they used undocumented VGA modes which could have cryptic bugs. Seems nothing's changed then ;)
- inDigiNeous 5y ago
  Well thats true :)