make sure _prereq etc are installed in the docker images

[dimpase]

This is a follow-up to #40460

Docker images need to get _prereqs installed, as noticed in #40460 (comment)

📝 Checklist

• The title is concise and informative.
• The description explains in detail what this PR is about.
• I have linked a relevant issue or discussion.
• I have created tests covering the changes.
• I have updated the documentation and checked the documentation preview.

⌛ Dependencies

[tobiasdiez]

Codewise looks good to me.

If you have the time, could you please apply the following patch, then those changes are automatically tested on github:

--- a/.github/workflows/push_to_docker_hub.yml
+++ b/.github/workflows/push_to_docker_hub.yml
@@ -1,10 +1,12 @@
 name: Build Docker images and push to DockerHub
 
 on:
+  pull_request:
+    paths:
+      - '.github/workflows/push_to_docker_hub.yml'
+      - 'docker/Dockerfile'
   workflow_dispatch:
     # Allow to run manually
-    branches:
-      - 'develop'
   push:
     tags:
       # Match all release tags including beta, rc

[dimpase]

@tobiasdiez - you can just push to the PR branch in my repo

[soehms]

I already started a run in my Repo: https://github.com/soehms/sage/actions/runs/17286731555. It gave:

#29 50.05 error: Sage source tree is unconfigured. Please run "./configure" first.
#29 50.05 note:  Type "./configure --help" to see the available configuration options.

[github-actions]

Documentation preview for this PR (built with commit f752004; changes) is ready! 🎉
This preview will update shortly after each push to this PR.

[soehms]

In the current log-file I see the same error-message that I've obeserved in this sage-devel thread. This has been fixed for a while and now shows up again.

BTW: This can also be seen in the log-file posted in #40708 (comment).

[dimpase]

In the current log-file I see the same error-message that I've obeserved in this sage-devel thread. This has been fixed for a while and now shows up again.

BTW: This can also be seen in the log-file posted in #40708 (comment).

sorry, it must be shell silliness (I don't have docker on a real machine to try out, so I need to see what CI says, just as in my undergrad years with punchcards jobs submitted to (a Soviet ripoff of) IBM-370 😄 )
Maybe I should try with podman.

[dimpase]

in the podman output, running podman build -f docker/Dockerfile --target sagemath-dev . I see

...
Setting up gfortran-11 (11.4.0-1ubuntu1~22.04.2) ...
Setting up g++ (4:11.2.0-1ubuntu1) ...
update-alternatives: using /usr/bin/g++ to provide /usr/bin/c++ (c++) in auto mode
update-alternatives: warning: skip creation of /usr/share/man/man1/c++.1.gz because associated file /usr/share/man/man1/g++.1.gz (of link group c++) doesn't exist
Setting up gfortran (4:11.2.0-1ubuntu1) ...
update-alternatives: using /usr/bin/gfortran to provide /usr/bin/f95 (f95) in auto mode
update-alternatives: warning: skip creation of /usr/share/man/man1/f95.1.gz because associated file /usr/share/man/man1/gfortran.1.gz (of link group f95) doesn't exist
update-alternatives: using /usr/bin/gfortran to provide /usr/bin/f77 (f77) in auto mode
update-alternatives: warning: skip creation of /usr/share/man/man1/f77.1.gz because associated file /usr/share/man/man1/gfortran.1.gz (of link group f77) doesn't exist
Processing triggers for libc-bin (2.35-0ubuntu3.10) ...
/bin/sh: 1: build/bin/sage-get-system-packages: not found

And this is because this apt-get is run on an empty image, without Sage files present. Duh...
This is sub-optimal. This means that every time _prereq and _bootstrap are changed, this has to change as well.

[dimpase]

I'll see if I can use docker's COPY to avoid needless git clone

[dimpase]

I'll see if I can use docker's COPY to avoid needless git clone

after wasting hours trying to figure out Docker's COPY, ADD, and RUN, I gave up and added missing deps directly.

[tobiasdiez]

Docker CI still fails, but I think that's because the last beta didn't built an image.

[soehms]

Docker CI still fails, but I think that's because the last beta didn't built an image.

Thank you for your effort. Perhaps we should try replacing the sagemath/sagemath-dev tags with our CI Docker images like sage-ubuntu-jammy-standard-with-targets to get rid of these targets in docker/Dockerfile. This would reduce maintenance overhead. I'll think about it after my vacation (i.e. after next two weeks).

[soehms]

Docker CI still fails, but I think that's because the last beta didn't built an image.

I suspect it's because Dima's repository doesn't have a suitable token for logging into DockerHub. The first job couldn't push the image needed by the second job.

I started another run from my repository, which should allow the login.

[dimpase]

Docker CI still fails, but I think that's because the last beta didn't built an image.

I suspect it's because Dima's repository doesn't have a suitable token for logging into DockerHub. The first job couldn't push the image needed by the second job.

I started another run from my repository, which should allow the login.

it's very confusing - why does one need to do here anything with sage's repo forks (e.g. mine of @soehms )?

[soehms]

Docker CI still fails, but I think that's because the last beta didn't built an image.

I suspect it's because Dima's repository doesn't have a suitable token for logging into DockerHub. The first job couldn't push the image needed by the second job.
I started another run from my repository, which should allow the login.

it's very confusing - why does one need to do here anything with sage's repo forks (e.g. mine of @soehms )?

I think this is because the workflow is inheriting credentials from the fork, since that's where the branch is located. However, we now have the Docker image on DockerHub:

https://hub.docker.com/layers/sagemath/sagemath/10.8.beta1/images/sha256-582c0d1dda26919a43208fad0ff3f0e8b995d0748671abd9ef4716e79593500a