Le 26/08/2019 à 00:08, Dima Pasechnik a écrit :
Vincent,
could you please try changing in spkg-install of eclib the line
--with-flint=$SAGE_FLINT_PREFIX \
to
--with-flint \
and see if this allows you to install eclib.
This worked! Thank you for the fix!
Vincent
--
You received this
I've opened https://trac.sagemath.org/ticket/28401 to deal with this.
Vincent,
could you please try changing in spkg-install of eclib the line
--with-flint=$SAGE_FLINT_PREFIX \
to
--with-flint \
and see if this allows you to install eclib.
Sorry for trouble,
Dima
Could you post your
On Sun, Aug 25, 2019 at 11:58 PM François Bissey wrote:
>
> Good catch. Yes, your explanation makes complete sense. Which means we have
> to make sure the right library path for pari is included in front of
> -lpari.
>
> Further examination may be required to figure out the presence of `-L/lib`.
On Sun, 25 Aug 2019 18:42 John Cremona, wrote:
> I think Vincent's solution is the correct one.
>
I guess I know how this made difference to running ./sage -i eclib. - but
it's not what you think.
Namely, in this case --with-pari option got processed before NTL and Flint,
and so the order of -L
Good catch. Yes, your explanation makes complete sense. Which means we have
to make sure the right library path for pari is included in front of
-lpari.
Further examination may be required to figure out the presence of `-L/lib`.
First it would be interesting to see if it appears in other packages
well, what goes wrong in the end is visible in eclib-20190226.log
specifically, in the failing ld call at the end there is
" -L/lib -L/opt/sage/local/lib -lflint -lntl -lpari "
which results in libpari picked up from /lib
I suppose /lib gets in the way just because Flint and NTL are coming
from
What would be interesting is the config.log for the eclib compile that goes
wrong.
And then I may have a few interesting commands to test further if needs be.
But extra configuration parameters in eclib’s spkg-install may well do the
trick.
François
> On 26/08/2019, at 3:42 AM, John Cremona
I think Vincent's solution is the correct one. If the sage install script
was like that it would only look where it is told and not in a standatd
place such as /usr/local. This should be done.
On Sun, 25 Aug 2019, 16:16 Vincent Delecroix, <20100.delecr...@gmail.com>
wrote:
> It becomes weird:
>
It becomes weird:
I untared eclib than ran
$ sage -sh
$ ./configure --with-pari=/opt/sage/local --with-boost="no"
--disable--allprogs
$ make
and everything went fine
Le 25/08/2019 à 16:47, Dima Pasechnik a écrit :
I suppose it is a eclib bug, which does not do linking in a "safe" way, but
I would be happy with the one from the system. But apparently, the one
shipped by archlinux is not good enough (have a look at the config.log
that I sent with my first e-mail).
Le 25/08/2019 à 16:48, Dima Pasechnik a écrit :
do you need to hack libpari, or you'd be happy with one from the
do you need to hack libpari, or you'd be happy with one from the system?
On Sun, 25 Aug 2019 15:10 Vincent Delecroix, <20100.delecr...@gmail.com>
wrote:
> I would be happier with a long term fix (and I don't want to uninstall
> my system pari as many other packages depend on it).
>
> Le
I suppose it is a eclib bug, which does not do linking in a "safe" way, but
relies on ld finding a libpari around the system in some chaotic way...
On Sun, 25 Aug 2019 15:10 Vincent Delecroix, <20100.delecr...@gmail.com>
wrote:
> I would be happier with a long term fix (and I don't want to
Ups, I wrote it wrong. His name is Grayson Jorgenson, not Jayson.
El viernes, 23 de agosto de 2019, 20:59:17 (UTC+2), mmarco escribió:
>
> Jayson Jorgenson implemented resolution of singularities for plane curves
> as a GSoC project a few years ago:
>
>
>
I would be happier with a long term fix (and I don't want to uninstall
my system pari as many other packages depend on it).
Le 25/08/2019 à 14:08, Dima Pasechnik a écrit :
it seems that eclib tries to link against the system's libpari, even though
it was meant to use Sage's libpari.
for a
it seems that eclib tries to link against the system's libpari, even though
it was meant to use Sage's libpari.
for a quick fix, either uninstall system pari, or uninstall Sage's pari,
and rebuild.
On Sun, 25 Aug 2019 15:01 Vincent Delecroix, <20100.delecr...@gmail.com>
wrote:
> The install log
The install log says
Successfully installed pari-2.11.1.p2
And I confirm that gp is installed (in /opt/sage/local/bin/gp)
and working correctly.
Vincent
Le 25/08/2019 à 13:46, François Bissey a écrit :
First question. Is sage’s pari installed?
François
On 25/08/2019, at 11:30 PM, Vincent
First question. Is sage’s pari installed?
François
> On 25/08/2019, at 11:30 PM, Vincent Delecroix <20100.delecr...@gmail.com>
> wrote:
>
> Dear all,
>
> I obtain an error while building eclib. The end of the log
> says
>
> /usr/bin/ld: avma: TLS definition in /lib/libpari.so section .tbss
>
> IIRC, singular returns the output of the resolution in a complicated way
> (defining several rings, and some ideals inside them). Maybe it would make
> sense to use the Sage schemes infrastructure to wrap that in a more
> user-friendly way (although the schemes code itself could use some
18 matches
Mail list logo