Hi,
Le 05/02/2015 01:50, Laurent Fousse a écrit :
On Wed, Feb 4, 2015 at 11:20 AM, Julien Puydt <julien.pu...@laposte.net>
wrote:
Hi,
Le 04/02/2015 20:04, Laurent Fousse a écrit :
Hi,
On Wed, Feb 4, 2015 at 10:46 AM, Julien Puydt <julien.pu...@laposte.net>
wrote:
Package: libecm-dev
Version: 6.4.4-2
While working on sagemath-on-debian, I met the following issue : the
sample code below, when compiled against sage's ECM tells me 1 is a
factor
of 1 ; and "No factor found" when compiled against debian's...
If anything, I would say there's a bug in sage's ECM. 1 is not a proper
factor of 1, and ECM is not about finding trivial factors.
Laurent.
what puzzles me is that they use ECM 6.4.4 too, and don't seem to patch
it... but still get a different result!
I haven't tried SAGE's version of ecm, but I can't reproduce your result on
my machine (amd64) using your code.
Uh... I wrote and tested that code on an amd64 box. What does it give you?
The ecm binary has a distinct error
code for "I found the input number", which it correctly returns for me when
it founds 1 as a factor of 1. I don't think the library function has such a
distinction, so you could argue that returning "success" when finding a
non-proper factor is correct, but I think that's something that should be
reported to upstream to clarify the documentation.
The fact that the same upstream 6.4.4 can give different results when
compiled by debian and when compiled by sage is a problem : there is
something fishy, and we should try to find more about it before we
contact upstream. Especially if you can't reproduce anything with my
minimal example code :-/
Snark on #debian-science
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
