(I’m failing to build a working binary) Please provide a snapcraft package/container as now all but one Linux distributions no longer package Sagemath.

[Laël Cellier]

Even by following this guide on a fresh minimal ᴠᴍ, whether it’s on red hat or Debian or Ubuntu, the resulting sagelib doesn’t load :

I’m meaning simply typing

from sage import *

on a python prompt triggers a segfault.

Of course this means building docs fails. Here’s the logs attached with lot of warnings.

[Nathan Dunfield]

I recommend you install SageMath via Conda:

https://doc.sagemath.org/html/en/installation/conda.html#install-from-conda-forge

In my experience, this method "just works" even on seriously old and cranky Linux distros.

Best,

Nathan

[Laël Cellier]

Except the latest conda binary version is sage 10.3 not 10.4 which lack the optional order= parameter on the log function on pari’s ffelt implementation.

In my case, the finite field’s order is very hard to factor so it’s impossible to solve my problem in old version of sagemath which is why I can’t just install an old package.

Sincerely,

[Dima Pasechnik]

Please post your config.log

CertIainly Sage does build and run on all sorts of Linux systems. But it's getting more and more bloated to support macOS, and this gets in the way
of Linux.

Dima

[Marc Culler]

Dear Laël,

On Sunday, August 25, 2024 at 2:29:44 AM UTC-5 "Dima" wrote:

CertIainly Sage does build and run on all sorts of Linux systems. But it's getting more and more bloated to support macOS, and this gets in the way of Linux.

It is true that Sage builds on linux, of course.  It is false that Sage bloat (real or perceived) is related in any way to the fact that Sage supports the macOS operating system.  The core issue is that parts of the Sage community have a very hard time accepting the idea that most Sage users are more interested in doing math than in compiling code, and would much prefer to be able to easily install a binary package which they can just use to having to build their own.  The fact that it is now possible for macOS users to do that seems to make macOS the target of belligerent attacks from time to time.

I have considered trying to build a binary package for linux, but I am currently leaning towards using AppImage rather than Snap.  If you have experience relevant to choosing between those I would be interested in hearing about it (off-list, of course).

- Marc

On Sunday, August 25, 2024 at 2:29:44 AM UTC-5 "Dima" wrote:

Please post your config.log

CertIainly Sage does build and run on all sorts of Linux systems. But it's getting more and more bloated to support macOS, and this gets in the way
of Linux.

Dima

On 24 August 2024 09:54:53 BST, "Laël" wrote:

Except the latest conda binary version is sage 10.3 not 10.4 which lack the optional order= parameter on the log function on pari’s ffelt implementation.

In my case, the finite field’s order is very hard to factor so it’s impossible to solve my problem in old version of sagemath which is why I can’t just install an old package.

Sincerely,

Le vendredi 23 août 2024 à 05:37:22 UTC+2, Nathan Dunfield a écrit :

I recommend you install SageMath via Conda:

https://doc.sagemath.org/html/en/installation/conda.html#install-from-conda-forge

In my experience, this method "just works" even on seriously old and cranky Linux distros.

Best,

Nathan

On Thursday, August 22, 2024 at 12:32:08 PM UTC-5 "Laël" wrote:

Even by following this guide on a fresh minimal ᴠᴍ, whether it’s on red hat or Debian or Ubuntu, the resulting sagelib doesn’t load :

I’m meaning simply typing

from sage import *

on a python prompt triggers a segfault.

Of course this means building docs fails. Here’s the logs attached with lot of warnings.

x

[Dima Pasechnik]

On 25 August 2024 15:33:41 BST, Marc Culler <marc....@gmail.com> wrote:
>Dear Laël,
>
>On Sunday, August 25, 2024 at 2:29:44 AM UTC-5 "Dima" wrote:
>
>CertIainly Sage does build and run on all sorts of Linux systems. But it's
>getting more and more bloated to support macOS, and this gets in the way of
>Linux.
>
>
>It is true that Sage builds on linux, of course. It is false that Sage
>bloat (real or perceived) is related in any way to the fact that Sage
>supports the macOS operating system. The core issue is that parts of the
>Sage community have a very hard time accepting the idea that most Sage
>users are more interested in doing math than in compiling code, and would
>much prefer to be able to easily install a binary package which they can
>just use to having to build their own. The fact that it is now possible
>for macOS users to do that seems to make macOS the target of belligerent
>attacks from time to time.

I don't agree with "having very hard time" etc.

It's just not so - the problem is that the current state of Sage (the bloat, related to the rather complicated macOS development model, coupled with hesitation to properly package Sage for macOS Homebrew) makes it very hard to package on a typical Linux distro, so that its users can easily install a binary package. That such efforts are stalled on Debian/Ubuntu and Fedora is very telling.

> The fact that it is now possible
>for macOS users to do that seems to make macOS the target of belligerent
>attacks from time to time.

Well, basically Sage the distro often runs in the way of sagelib (the only component
really needed).
Attacks? No, it's merely pointing out the disregard of needs of Linux packagers by a number of Sage developers.

As long as this is not sorted out, the future of Sage as the multiplatform project looks bleak.

Last but not the least, you continue to ignore the need to install sagelib into existing Python venv,
the need that arises whenever anyone doing maths computations with Python merely wants to use sagelib.

For instance the latter is how one would typically use SciPy.
And SciPy gets about 2000 citations in refereed publications per year, without offering any kind of SciPy the distro.
SageMath gets about 10 or 20, despite, as you claim,
being much more user-friendly.

Dima

><https://sagemanifolds.obspm.fr/install_ubuntu.html> on a fresh minimal ᴠᴍ,