On Thu, Dec 6, 2012 at 2:22 PM, Roy Stogner <[email protected]>wrote:
> My general rule is not to do self-testing in opt mode when that > testing has any time or memory cost. > > Now that you got me to think about it: in this case the costs are "a > handful of operations and one word of memory per EquationSystems > object" and so my general rule is probably insanely wrong; testing > every time is the way to go. > Yeah - we default to testing and throwing errors in any non-compute intensive code sections. Any check that is only run once for instance... even if it might be a heinous thing in and of itself. Our solves are so huge that we can basically get away with _anything_ once. ;-) We're a bit closer to "code users" (ie, non-code developers) than you guys are though... so that policy might make more sense for us. Derek ------------------------------------------------------------------------------ LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial Remotely access PCs and mobile devices and provide instant support Improve your efficiency, and focus on delivering more value-add services Discover what IT Professionals Know. Rescue delivers http://p.sf.net/sfu/logmein_12329d2d _______________________________________________ Libmesh-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/libmesh-users
