On 08/17/10 03:31 AM, Bill Hart wrote:
I have no sympathy for anyone who wants to include that code in Sage.
I have very little myself.
I know John Cremona has made the point SYMPOW is much faster than anything he
has written at computing the modular degree of an elliptic curve, but I'd
personally not want to trust this code.
It obviously works in some cases, and works faster than anything else available.
But how reliable is it? That's what concerns me.
Of course we can never 100% trust software, but there are indicators here to
make me *much* more mistrustful than usual.
No documentation.
Well the README file has some (very minimal) documentation.
Comments in the code are far from adequate.
It's virtually obfuscated.
It can't however be entered in to The International Obfuscated C Code Contest,
http://www.ioccc.org/
since it's not legal C! Using the return value of functions declared as void is
not permitted in C, thought gcc rather stupidly accepts it.
Test code is where?
Exactly. How well has it been tested?
It seems to be that it difficult to test it, as there's nothing else able to
compute some of these things quickly.
At any rate I used my "special powers" to divine that the following
files are required to get the modular degree functions working:
moddeg.c QD.c eulerfactors.c conductors.c compute.c ec_ap.c util.c
factor.c ec_ap_bsgs.c ec_ap_large.c compute2.c analytic.c sympow.h
Does that make your "porting" job any easier?
Thanks for your efforts Bill, but you need to take a step back and try to
convince me *why* I should bother "porting" this.
IMHO, this is not a portability problem, but a quality control problem.
Perhaps William will look at this for the Cygwin port, as it is also a package
breaking on Cygwin.
SYMPOW is not something I feel willing or able to do.
Bill.
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
