> Compilation performance is not generally a problem, although anything that takes > longer than 10 seconds to build when trying to fix a bug is annoying. If the software > takes ages to get to the point of the problem then you'll be tearing your hair out.
Speak for yourself; our codebase takes 45+ minutes to build from scratch even on fast machines and a no-op build is 2 minutes (crappy recursive makefiles for the win!). However we have ways of mitigating this; developers don't usually have to build everything themselves. > When trying to debug a problem that only occurs when reading a gig of xml and > inserting it into a database, I'll take all the CPU & disk performance that I can > get.
I would humbly suggest that whoever decided to represent a gigabyte of information in XML should be ... re-educated, to put it politely. I personally will never use XML for anything given any possible alternative, but that's just me. Well, I lie; if I'm going to MARK UP some TEXT I might use the eXtensible Markup Language, but for representation of structured data, never.
|