What Every Computer Scientist Should Know About Floating-Point Arithmetic
docs.oracle.com
docs.oracle.com
http://news.ycombinator.com/item?id=453396
http://news.ycombinator.com/item?id=687604
http://news.ycombinator.com/item?id=1746797
http://news.ycombinator.com/item?id=1937182
http://news.ycombinator.com/item?id=1982332 <-- comments
http://news.ycombinator.com/item?id=2051062
http://news.ycombinator.com/item?id=2287497
http://news.ycombinator.com/item?id=2614330
http://news.ycombinator.com/item?id=2684497
http://news.ycombinator.com/item?id=3058560
http://news.ycombinator.com/item?id=3428112
You have been very helpful on keeping me on my toes. I'll be working on a tool to prevent me from making such mistakes again.
Problem is that from some points of view it's not really a mistake. We are encouraged to submit things we find that are expected to be of interest to hackers, and this certainly should be. The lack of a comprehensive duplicate detector and the fact that items do fade into obscurity together mean that repeats like this are inevitable, and not necessarily "wrong."
I'd love to see a solution to this niggle. Others say it's not a problem, and I've even had some rather unpleasant emails over the issue, but I maintain that repeats like this show the need for a companion site with "Classic News" (whatever that may mean)
I started to develop one, but life took over, and I don't have the skills to knock up web prototypes quickly, so it languishes, unloved. Maybe one day something will be done to create a resource that preserves the knowledge and judgment implicit in these resubmissions.