Systems Administration Ideology Discussion
IMO, "leave the problem in place to bite someone later" is the worst possible response. The system may be running, but who knows what scripts are going to fail in surprising ways?
-- jfm
Oh, it's a horrible response! Admittedly I wrote this post on a train nearly a year ago, so my thinking may have changed somewhat but two things.
More often than not, this is what happens, ideal response or not. Known problems with systems aren't fixed until they present an actual problem (or there's some sort of regulation/policy requirement.)
I think "deal with potential problems when they're actually problems," is probably the best way to avoid over-engineering. Better, I think to get something into production than to spend significant time developing the theoretically perfect design and implementation only to have misjudged the problem, demand, or stresses of the system.
I mean this is what the "systems are disposable" aspect of cloud computing does. Why make things perfect when you can make them good enough for now? Obviously there are times and products where "good enough for now," is technically perfect and impeccably implemented, but those situations emerge pretty quickly.
Or am I really crazy?
-- tychoish
No, you're right-ish. 1. is bad, but possibly inevitable without a strong culture of "fix it now" (or, at least, put it in the bug tracker now). 2. is a restatement of YAGNI. It's just the example is bad, because in this case, you are bloody well GNI.
-- jfm