From: Jacob Martin Bohn Lorensen (jablo@diku.dk)
Date: 11/04/92


From: jablo@diku.dk (Jacob Martin Bohn Lorensen)
Subject: Re: Pari-1.37 for Linux
Date: 4 Nov 1992 20:18:33 GMT

ed@orca.wimsey.bc.ca (Ed Rodda) writes:

>If anybody is interested in running pari on their Linux box, they can get it
>from <math.ucla.edu> in pub/pari. It's easy to find. It compiles easily,
>no warnings, only one small glitch. I compiled it for an ascii terminal. If
>you want the X-windows version, there is a change to the Makefile. I don't
>use X-windows, but I think it will compile ok.

>Pari is a sophisticated math solver, very fast.

>Hope this isn't old news.

Well, it wasn't for me. I ftp'ed it and tried it out on my newly
compiled 0.98.3 (+ fs/inode.c patch) just to test the new math emulater.
Uh oh. On the test file, it gives a "Floating point exception" and
dumps core quite early in the tests.

For example, (from memory, not from test file)
  ? roots(x^1+x^2+x^3+x^4)
gives Floating point exception and dumps core.

On the 0.98.1 kernel it works as you say.

>Ed

BTW - does anybody know, if it is possible to alltogether drop
/lib/libsoft (forgot the exact name) and use the /lib/libhard (ditto)
library with the new math emulator?

Mange Hilsener,
  jacob.

-- 
No signature yet !