| Subject: | Request new builtin numeric functions |
| Date: | Thu, 20 May 2010 18:45:37 +0200 |
| To: | farnsworth-bugs [...] googlegroups.com |
| From: | Zsbán Ambrus <ambrus [...] math.bme.hu> |
I'd like to request that you add the following builtin numeric
functions to farnsworth. Each of these should be trivial to implement
because you just have to wrap a perl or pari function.
1. ldexp, fmod - these would be just straight wrappers over the
numeric functions which can be found in the perl module POSIX acting
on machine floats.
2. frexp, modf - similar, except either they have to return an array
of two numbers (inconvenient because you don't have list assignment
right now), or you have make them two functions each.
3. ffloor, fceil, frint, ftrunc - similar, these would return a float
instead of a very large bigint for large numbers, while the floor etc
functions would keep returning bigints.
4. div, mod, quo, rem - integer division:
div and mod are mathematician's division, they round quotient
towards minus inf so x mod y has same sign as y,
quo and rem are computer scientist's division, they round quotient
towards zero so x rem y has same sign as x.
these should be builtins calling the pari functions so they work
correctly (and fast) on bigints.
Ambrus