Radix sort/Talk

THANK YOU FOR WRITING THIS WHOEVER DID!!!

---

Does the sample implementation given depend on the endianness used by the computer? I'm not quite sure what exactly the >> operator does. --AxelBoldt

The value of x >> y is floor(x / 2^y), provided x and y are non-negative, and y is not too big. So it doesn't depend on endianness in this case. But if x is negative, then the value is undefined, so this is a problem. Another problem with the code in the article is that it assumes that CHAR_BIT is 8. --Zundark, 2001 Dec 11

So maybe we should make them unsigned ints to take care of the sign issue? Is there a clean way to deal with the CHAR_BIT thingy?

I think we should probably have an implementation that deals with keys that are (arbitrary length) strings, since that seems to be the more interesting case. --AxelBoldt

I changed the int array to unsigned. The CHAR_BIT thing is more problematic, so I just added a note pointing it out. On most compilers for desktop computers CHAR_BIT is 8, so this isn't too bad a restriction. --Zundark, 2001 Dec 11