MR52: upgrade ubuntu system in docker image

[sagetrac-gitlab-bot]

Julian Rüth (@saraedum) opened a merge request at https://gitlab.com/sagemath/sage/-/merge_requests/52:

---

this fixes an issue with line endings in CI. The old version of
git on ubuntu xenial does not understand our .gitattributes or it does
not understand the .git directory that we copy into the docker build
context from a much newer version of git. Anyway, the broken line
endings are gone now.

Component: docker

Author: Julian Rüth

Branch/Commit: 0a96413

Reviewer: Frédéric Chapoton

Issue created by migration from https://trac.sagemath.org/ticket/31271

[fchapoton]

comment:1

still fails on gitlab

here: https://gitlab.com/sagemath/dev/trac/-/pipelines/244687774

[fchapoton]

comment:2

maybe because latest is stuck at 9.0, though

[saraedum]

comment:3

The CI run for this branch is here: https://gitlab.com/sagemath/dev/sage/-/jobs/976444339

It seems to work since git status --porcelain does not report any pending changes anymore.

[saraedum]

comment:4

While this fixes one problem, we now have a [dochtml] Illegal instruction (core dumped) in another CI step.

[saraedum]

comment:5

Probably happens because the image was built on Skylake and the tests ran on Sandy Bridge.

[sagetrac-gitlab-bot]

comment:6

New commits added to merge request. I updated the commit SHA-1. New commits:

07fb9bf

build docker image without CPU specific flags

[sagetrac-gitlab-bot]

Changed commit from 2756453 to 07fb9bf

[fchapoton]

comment:7

so needs review ?

[mkoeppe]

comment:8

You already had ENV SAGE_FAT_BINARY yes, but you might be running into #29537 here

[sagetrac-gitlab-bot]

comment:9

New commits added to merge request. I updated the commit SHA-1. New commits:

0a96413

Revert "build docker image without CPU specific flags"

[sagetrac-gitlab-bot]

Changed commit from 07fb9bf to 0a96413

[saraedum]

comment:10

Replying to @fchapoton:

maybe because latest is stuck at 9.0, though

It fails on gitlab because the sagemath-dev:develop image is broken. (due to this very issue.)

I ran it in a separate pipeline and it seemed to work.

[saraedum]

comment:11

Let's not worry about the SAGE_FAT_BINARY issue here. That's a different problem.

[fchapoton]

comment:12

ok, let's move on

[fchapoton]

Reviewer: Frédéric Chapoton

[vbraun]

Changed branch from u/galois/mrs/52/docker-attributes to 0a96413