Thursday, July 7, 2011

Subversion woes pt2

Okay, In fairness to subversion, I feel the need to amend my past statement. Subversion doesn't need to die, but all its poor implementations do. I finally got decent performance out of it, after removing Tortoise (Not even a Hoveraroud would have saved that one), Installing Ankh and a command line SVN tool. Both of which raised my times from, O(N) where N = Files in source, to O(N) where N = files checked out, or files modified since the last sync.

Which means, I ended up installing 3 tools, to get what perforce does with 1. On second thought, subversion can still diaf.

Wednesday, July 6, 2011

Yes I'm rewriting THAT system before beta

Every now and again in the natural evolution of a project. You are going to end up with 2 similar paths for doing task X. Like all good divergent paths, eventually path A != B. You end up spending days, and days making A behave like B. But A isn't B. Too many times I watch people say, duct tape it and move on. Fix it, it costs less. If you don't have a producer that can understand frontloaded time now, saves bugs later during live, I suggest practicing in a mirror, your spouse, my cat (he has plenty of experience listening to me ramble, often sober).

That's why with a beta looming in the future, I'm re-writing a major gameplay system from scratch. We really don't spend enough time as engineers paying down the technical debt we accrue in the "Just ship this milestone" mentality. For Console guys, it's not as bad, its often in the gameplay layer and we can in practice, dispose of those parts. For live teams, that debt becomes your successors thorn.