Well my new PC from Mesh will be a few weeks. I opted for a 256 MB Video card, 2.8 gig pentium and no monitor (my lovely iiyama flat panel will stay ;))
Been mostly biug fixing at work. Ive been trying to do this zero-defect software development thing, which is more an attitude than anything else. You find some very obscure bugs that way, and get more and more confident about the stability of your code.
Ive just been discussing with work people how it should never be needed for people to work real late the night before a demo/release/milestone etc.
I always try and liken this kind of thing to launching a spaceship or a missile. You wouldnt fiddle with the code of space shuttle 10 hours before takeoff(although judging by recent events it seems that way). You get to a point, WAY before release, where there is a lock on features, and you make damn sure that every bug fix after that point doesnt introduce any more bugs.
The tendency amongst coders is to put in a quick bodge fix, once they have spotted the problem, without reallky analysing what went wrong, and what the ramifications are.
Personally I reckon that in the last month of a project, ANY code change shousl be looked over and approved by at least 1 other coder, and there is a good argument for all coders to work in pairs, doing end-of-week brief code reviews of each others work.
Of course this doesnt happen much in game dev, but methinks it should.
Been mostly biug fixing at work. Ive been trying to do this zero-defect software development thing, which is more an attitude than anything else. You find some very obscure bugs that way, and get more and more confident about the stability of your code.
Ive just been discussing with work people how it should never be needed for people to work real late the night before a demo/release/milestone etc.
I always try and liken this kind of thing to launching a spaceship or a missile. You wouldnt fiddle with the code of space shuttle 10 hours before takeoff(although judging by recent events it seems that way). You get to a point, WAY before release, where there is a lock on features, and you make damn sure that every bug fix after that point doesnt introduce any more bugs.
The tendency amongst coders is to put in a quick bodge fix, once they have spotted the problem, without reallky analysing what went wrong, and what the ramifications are.
Personally I reckon that in the last month of a project, ANY code change shousl be looked over and approved by at least 1 other coder, and there is a good argument for all coders to work in pairs, doing end-of-week brief code reviews of each others work.
Of course this doesnt happen much in game dev, but methinks it should.
posted by cliff at
7:16 PM
