On 08/23/10 01:46 PM, Bill Hart wrote:
Yes, I wrote to Mark. I asked him for permission to post his response
in toto here. But he did not reply to that email. His response was
*very* helpful and I suspect anyone wanting to reimplement Sympow
would do well to contact Mark directly and also John Cremona.
The two important points I got from Mark's reply were:
1) Mark is definitely *not* intending to "maintain" Sympow as a
software project. He "maintains" it primarily for his own use, and if
it runs on his machine that is enough for him.
2) The modular degree stuff, in the range that it is feasible, would
probably work just fine if we ditched the QD stuff and just replaced
it with doubles. If this is all we really require from Sympow, it
makes good sense to do this. Long doubles are evil!
Bill.
From my own perspective, the code is so convoluted that I would not want to try
to replace anything.
In any case, it now works. Mike has tested it on Cygwin, and I've tested it on
Solaris, OpenSolaris, Linux and OS X.
I'm still amazed the code works at all. It's hard to know what one is supposed
to do with code which has
void foobar()
main() {
return foobar()
}
I think its a bug in gcc in compiling this, but my attempts to produce a subset
of it which I can report ot the gcc developers has failed.
The Sun C compiler naturally rejects this code, as it's not C.
When you see things like a configure script that starts
#! /bin/sh
then has code to test if 'sh' exists or not, I think you can understand why I am
reluctant to do any more with SYMPOW than I absolutely need to. Replacing it
with double, long double, mpfr is not something I want to do.
$ rm -rf sympow
would suite me better, but I think the patch I have works.
Dave
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
