dodona-edu / docker-images Goto Github PK

• dodona-anaconda3 to Python 3.6
• dodona-bash to Python 3.6 and ImageMagick 7.0.4-10

@pdawyndt and @nineiryn: could you mark above if you've tested on Naos? (Note: Naos seems very slow, timeouts could occur at random.)

Original issue by @ninewise on Mon Feb 20 2017 at 14:11.
Closed by @ninewise on Mon Apr 24 2017 at 18:07.

I can't take the new worker server into production because building the back images fails on imagemagick not being available.

--2017-11-30 15:24:37--  http://www.imagemagick.org/download/ImageMagick-7.0.7-8.tar.gz
Resolving www.imagemagick.org (www.imagemagick.org)... 198.72.81.86
Connecting to www.imagemagick.org (www.imagemagick.org)|198.72.81.86|:80... connected.
HTTP request sent, awaiting response... 404 Not Found
2017-11-30 15:24:37 ERROR 404: Not Found.

@dmouriss @dmouriss Why don't we install this with apt-get?

Original issue by @bmesuere on Thu Nov 30 2017 at 16:34.
Closed by @bmesuere on Thu Nov 30 2017 at 16:44.

@dmouriss the goals of this repository was to bundle the images that are needed for the workers/judges. I think the tutor is important enough to be moved to its own repository.

Original issue by @bmesuere on Wed Sep 14 2016 at 20:50.
Closed by @dmouriss on Tue Oct 04 2016 at 16:08.

In commit #244a381 many R packages were added to the R docker, one of which being "topicmodels".
However, there occurred an error while installing this package:
ctm.c:29:10: fatal error: gsl/gsl_rng.h: No such file or directory
line 16855 in the R Publish to Registry section (here)

While looking up this error message I came across someone experiencing a silimar issue on this blogpost and this stackoverflow post.
There seems to be an issue surrounding the installation of GSL (The GNU Scientific Library).
I am out of my depth when it comes to implementing this fix on docker.
Could someone please aid me in solving this issue to we can use the topicmodels package in our Dodona exercises using the R-judge?

From https://github.ugent.be/dodona/dodona/issues/389

Importing larger modules like numpy, scipy, pandas takes some time if they haven't been imported before (Python stores compiled versions on disk). Since these modules have not been imported before in the docker instance, each time it gets started, the import of the modules gives some overhead. Could be bybpassed by importing some of these modules when building the docker.

Original issue by @charvp on Thu Sep 27 2018 at 14:36.

The C# image doesn't adhere to docker file best practices and doesn't pass our linting. While this might look innocent, it can have unexpected consequences: images built at a different time on a different worker node can behave differently.

To fix this, the following changes need to be made:

dodona-csharp.dockerfile:1 DL3006 Always tag the version of an image explicitly
dodona-csharp.dockerfile:4 DL3009 Delete the apt-get lists after installing something
dodona-csharp.dockerfile:5 DL3008 Pin versions in apt get install. Instead of `apt-get install <package>` use `apt-get install <package>=<version>`
dodona-csharp.dockerfile:5 DL3015 Avoid additional packages by specifying `--no-install-recommends`
dodona-csharp.dockerfile:6 DL3008 Pin versions in apt get install. Instead of `apt-get install <package>` use `apt-get install <package>=<version>`
dodona-csharp.dockerfile:6 DL3015 Avoid additional packages by specifying `--no-install-recommends`
dodona-csharp.dockerfile:19 DL3002 Last USER should not be root

Inspiration on fixing these can be found at #13 and at https://github.com/hadolint/hadolint/wiki.

Line 22 in r.dockerfile adds the 'car' package to the list

But when I tried loading the library in an dodona exercise, it have me an error:
Error while evaluating context: there is no package called ‘car’

It seems like there goes something wrong installing the package, so when the package is called, it is not available.

Exercise I am referring to:
https://dodona.ugent.be/nl/courses/339/series/3514/activities/635124618/

The R docker image I'm preparing, is based on alpine linux and doesn't come with a bash shell (alpine uses the ash shell).

Can #!/bin/bash be replaced with #!/bin/sh in main.sh and logger.sh? If not, I'll install the bash shell in the docker image (but I like to keep it as minimal as possible).

Original issue by jfmeys on Thu Apr 05 2018 at 17:55.

User request to support numpy for Python (submission).

The prolog image doesn't adhere to docker file best practices and doesn't pass our linting. While this might look innocent, it can have unexpected consequences: images built at a different time on a different worker node can behave differently.

To fix this, the following changes need to be made:

dodona-prolog.dockerfile:4 DL3009 Delete the apt-get lists after installing something
dodona-prolog.dockerfile:5 DL3008 Pin versions in apt get install. Instead of `apt-get install <package>` use `apt-get install <package>=<version>`
dodona-prolog.dockerfile:5 DL3015 Avoid additional packages by specifying `--no-install-recommends`
dodona-prolog.dockerfile:24 DL3002 Last USER should not be root

Inspiration on fixing these can be found at #13 and at https://github.com/hadolint/hadolint/wiki.

The Haskell image doesn't adhere to docker file best practices and doesn't pass our linting. While this might look innocent, it can have unexpected consequences: images built at a different time on a different worker node can behave differently.

To fix this, the following changes need to be made:

dodona-haskell.dockerfile:4 DL3009 Delete the apt-get lists after installing something
dodona-haskell.dockerfile:5 DL3008 Pin versions in apt get install. Instead of `apt-get install <package>` use `apt-get install <package>=<version>`
dodona-haskell.dockerfile:5 DL3015 Avoid additional packages by specifying `--no-install-recommends`
dodona-haskell.dockerfile:6 DL3008 Pin versions in apt get install. Instead of `apt-get install <package>` use `apt-get install <package>=<version>`
dodona-haskell.dockerfile:6 DL3015 Avoid additional packages by specifying `--no-install-recommends`
dodona-haskell.dockerfile:33 DL3002 Last USER should not be root

Inspiration on fixing these can be found at #13 and at https://github.com/hadolint/hadolint/wiki.