François Bissey wrote:
> 7.0.3 over there:
> https://gforge.inria.fr/frs/?group_id=135
> fixes stuff for gcc 5+ and gmp 6.1.0 and ppc64.
See ticket, I pushed a branch (can be reviewed / tested).
-leif
> Will try on recent OS X to check if that bit is fixed.
> I was thinking of another QA proble
7.0.3 over there:
https://gforge.inria.fr/frs/?group_id=135
fixes stuff for gcc 5+ and gmp 6.1.0 and ppc64.
Will try on recent OS X to check if that bit is fixed.
I was thinking of another QA problem.
Francois
On 03/08/16 12:48, leif wrote:
Francois Bissey wrote:
https://trac.sagemath.org/tic
Nice. GNU Savannah in first... what about autoevaluation? ;-P
On 02/08/16 19:30, leif wrote:
[For some reason, creating a new thread via gmane doesn't work -- or
there will probably come a duplicate later.]
I guess that if we start rating googlegroups with respect to the same
criteria its gra
Francois Bissey wrote:
> https://trac.sagemath.org/ticket/20385
> for upgrade to 7.0.
> Patches need to be ported to the new version.
Sage currently only has this poor patch:
commit 1f3e8dc1a1f72f99c8c758c80463138bbef90534
Author: John H. Palmieri
Date: Thu Sep 24 07:43:14 2015 -0700
trac
[For some reason, creating a new thread via gmane doesn't work -- or
there will probably come a duplicate later.]
The FSF has created criteria for evaluating repository hosters w.r.t.
ethics:
https://libreplanet.org/wiki/Repository_Ethics
They've also ranked GitHub (follow the second link [1]
https://trac.sagemath.org/ticket/20385
for upgrade to 7.0.
Patches need to be ported to the new version.
If there is 7.1 is our=t lets move to that.
François
> On 3/08/2016, at 11:14, leif wrote:
>
> leif wrote:
>> I'd say it's an upstream bug that whether ECM considers 1 a factor of 1
>> depe
leif wrote:
> I'd say it's an upstream bug that whether ECM considers 1 a factor of 1
> depends on the underlying library (GMP 5.x or MPIR vs. GMP 6.x).
>
> But the last release of GMP-ECM (6.4.4) dates back nearly three and a
> half years... (There was some activity to make a 7.x release years a
leif wrote:
> Francois Bissey wrote:
>> I would have to check my discussions with Paul but I think that’s why there
>> is a difference for the ecm doctest (with Paul saying that he likes the
>> current
>> gmp behaviour).
>
> We fixed FLINT w.r.t. this (invertability modulo +/-1) at least, such
>
Jori Mäntysalo wrote:
> On Sat, 30 Jul 2016, leif wrote:
>
>> Well, in addition a (boolean) 'certify' (or 'certificate', more
>> misleading here) parameter/keyword could just mean that you don't want a
>> /probabilistic/ result, as in for example is_prime(), where it's called
>> 'proof' (not even
Francois Bissey wrote:
> I would have to check my discussions with Paul but I think that’s why there
> is a difference for the ecm doctest (with Paul saying that he likes the
> current
> gmp behaviour).
We fixed FLINT w.r.t. this (invertability modulo +/-1) at least, such
that FLINT > 2.4.5 copes
On 2 August 2016 at 15:40, Nils Bruin wrote:
> On Tuesday, August 2, 2016 at 3:09:48 AM UTC-7, Jori Mäntysalo wrote:
>>
>>
>> For primality certificate I just learnt something new. Thanks!
>>
> A non-primality certificate is usually even cheaper: it's usually a number a
> such that a^n-1 is not di
On Tuesday, August 2, 2016 at 3:09:48 AM UTC-7, Jori Mäntysalo wrote:
>
>
> For primality certificate I just learnt something new. Thanks!
>
> A non-primality certificate is usually even cheaper: it's usually a number
a such that a^n-1 is not divisible by n (and a itself not divisible by n);
see
On Sat, 30 Jul 2016, leif wrote:
Well, in addition a (boolean) 'certify' (or 'certificate', more
misleading here) parameter/keyword could just mean that you don't want a
/probabilistic/ result, as in for example is_prime(), where it's called
'proof' (not even 'prove' or 'provably').
True. So w
On Sat, 30 Jul 2016, Volker Braun wrote:
That just illustrates my point that its a bad api choice; Without
consulting the docs its impossible to know if and what a certificate is
going to mean. And if one has to look into the docs anyways then we
could just point users at a better-named other
I would have to check my discussions with Paul but I think that’s why there
is a difference for the ecm doctest (with Paul saying that he likes the current
gmp behaviour).
François
> On 2/08/2016, at 21:35, Jean-Pierre Flori wrote:
>
> Also note that GMP changed its behavior for some gcd or mod
I think that I some point I wanted to support switching btw GMP and MPIR
and then someone convinced me it was hopeless.
The only supported solution should be to distclean, reconfigure, rebuild.
If that does not work, we should fix it.
As far as error are concerned, I remember that MPIR and GMP be
16 matches
Mail list logo
