[prev] [thread] [next] [lurker] [Date index for 2007/01/26]
It was thus said that the Great Yossi Kreinin once stated: > Some programs crash when they crash. Can't their authors handle errors? > > By "errors", I mean things like *(int*)0xACDC = 0xABBA; > By "handle", I mean things like signal(SIGSEGV, SIG_IGN); > > So the program won't crash when it crashes. Instead, the zombie program > will keep running from hell knows what point with hell knows what data > overwritten by hell knows what garbage with hell knows what consequences. > This is obviously the better behavior considering the breakdown of the user > population: Cow-orker of a friend stuff here, but ... A friend of mine loved to rant on the code he had to work with written by his cow-orker (who worked from home and lived in a state far from the office). It seems this cow-orker would write ... shall we say, less than stellar software and when it crashed (usually because of SIGSEGV), instead of tracking down the problem and fixing that, would catch the exception and try the operation again, using a different algorithm. So, if he was using a QuickSort and it bombed, he would catch that exception, and then try Bubble Sort. And if that failed, oh, I don't know, he would catch *that* exception and try something like BogoSort. And it wasn't just sorting mind you---it could be walking a graph, or trying to draw a window---all sorts of things. If it failed and it wasn't obvious what the problem was, another catch was added and the operation tried again. -spc (There's a reason why I never maintain code not written by me)
Generated at 23:01 on 06 Feb 2007 by mariachi 0.52