I have put an updated eclib at
http://www.warwick.ac.uk/staff/J.E.Cremona/eclib-20071231.p1.spkg
This version, which builds on mabshoff's eclib-20071231.p0.spkg,
handles the interface with NTL's ZZ_p class better (using a cached
list of ZZ_pContext's for those who know NTL) which should certailny
be a lot more efficient, and might even solve the problems people had
trying to compile this with gcc-4.3 (which I do not have).
Since other parts of Sage use the ZZ_p class it might be worth using
what I did here (in src/procs/gf.{h,cc}) elsewhere too. I have a
global object of type map<ZZ,ZZ_pContext> consisting of an in
initially empty list of pairs [p,c] where p is a prime and c is a
ZZ_pContext which stores the internal NTL data for working mod p.
When I need to switch to a new modulus, instead of just calling
ZZ_p::init() each time (which was happening a lot of times for the
same p in programs such as mwrank), I look to see if that p has a
saved context and if so just restore it. If not, is do ZZ_p::init and
then save the context. So each prime only gets init'ed once.
It is possible that this will have solved other mysterious problems,
since the NTL documention (see http://www.shoup.net/ntl/doc/ZZ_p.txt)
says
"One should also not presume that things will work properly
if the modulus is changed, but its value happens to be the same---
one should restore the same "context", from either a ZZ_pBak
or a ZZ_pContext object."
Up to now I *had* been making such a presumption.
It is still true that this spkg will not build correctly in Sage 2.9.1
(or even 2.9.2) since there is a small change to the interface which
requires a similar small change to the wrapping code, but this has
been put off at least until after the AMS meeting. But it should
build fine as a stand-alone package.
John
--
John Cremona
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-devel@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/
-~----------~----~----~----~------~----~------~--~---
