[07:25:36] <Chess>	 random question but how might a function like "fastinvsqrt" be dealt with in wikifunctions?
[15:10:47] <vrandecic>	 depends. If you want to implement the historic algorithm, you would need to create types that are amenable to that algorithm. So you could capture that. But that would be more of a historic curiosity.
[15:10:48] <vrandecic>	 What I would expect to happen is that there would be a function declaration for "inverse square root" that has several implementations. One of these implementations could be following the fastinvsqrt algorithm, but there would be other implementations as well. The system would decide, which implementation to run.
[17:29:28] <Chess>	 vrandecic: what level of inaccuracy is tolerated for an implementation before we tell it to get out of wikifunctions?
[17:48:56] <vrandecic>	 up to the community, and might depend case by case. Testers will be able to run with an epsilon as defined for the given function.