begin quoting Andrew Lentvorski as of Sat, Jun 14, 2008 at 11:33:09AM -0700: > I saw this today and kinda nodded vigorously. The issues he talks about > are the tendency of many computer folks to develop code "in a cave". > Basically, they want to go off, write lots of code, and then expect the > project to integrate the code bomb. > > http://blog.red-bean.com/sussman/?p=96
Yah, I read that. We call it "throwing the code over the wall", as it often happens at the end of a subcontract, and more often than not, everyone is in piss-on-the-fire mode and if it compiles and integrates, it's not looked at too closely. It's HARD to keep this from happening. And it's HARD to keep oneself from doing this if it's not a key theme. I've been playing with hg and git, and I'm finding that the ability to check-in changes locally is encouraging me to not commit to the main repository daily. (Not that CVS/perforce/etc centralized systems force anyone to do so; I've worked on teams that happily developed in caves using a centralized version control system. Of course, those programmers would also check in changes that would break the compile, which is worse than checking in buggy code.) -- There's what I know is right, and then there's what I actually do. Stewart Stremler -- [email protected] http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
