![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
I've fixed the J Mandelbrot set program. Needless to say, the fix took precisely three characters :-)
[By the way, did anyone read that post? Does anyone else think J is interesting? Or did you just see a wall of punctuation and get scared off?]
( The gory details )
Now here's the interesting bit: the Haskell version seems only to work because of what I can only assume is a bug in GHC's handling of IEEE 754 special values. The Haskell version behaves exactly like the J version until it gets to the stage of taking the magnitude of NaN + i NaN, at which point (in Hugs) it dies with an "arithmetic overflow!" error (grrrr: that's what Infinity and NaN are for), or (in GHC) it returns Infinity (which is greater than 2, of course). G'huh? How does Infinity make sense here? It's not a number, it doesn't have a magnitude, much less an infinite one. Even more weirdly,
Yet again, I suffer pain because someone didn't read IEEE 754 properly. Which is the IEEE's fault for charging heavily for their standard documents, rather than making them freely available like they would have done if they actually cared about wide dissemination and interoperability. Grrrrr.
[By the way, did anyone read that post? Does anyone else think J is interesting? Or did you just see a wall of punctuation and get scared off?]
( The gory details )
Now here's the interesting bit: the Haskell version seems only to work because of what I can only assume is a bug in GHC's handling of IEEE 754 special values. The Haskell version behaves exactly like the J version until it gets to the stage of taking the magnitude of NaN + i NaN, at which point (in Hugs) it dies with an "arithmetic overflow!" error (grrrr: that's what Infinity and NaN are for), or (in GHC) it returns Infinity (which is greater than 2, of course). G'huh? How does Infinity make sense here? It's not a number, it doesn't have a magnitude, much less an infinite one. Even more weirdly,
sqrt (nan*nan + (nan*nan)) returns NaN as expected.Yet again, I suffer pain because someone didn't read IEEE 754 properly. Which is the IEEE's fault for charging heavily for their standard documents, rather than making them freely available like they would have done if they actually cared about wide dissemination and interoperability. Grrrrr.
Tags:
