ioc: install busybox symbolic links.

[henriquesimoes]

Make all busybox goodies available in the PATH without having to worry about them being from busybox. This is done automatically for Alpine, but not for Debian.

This removes the need to install wget, since it is already provided by busybox. However, its version packaged by Debian does not implement -o flag, requiring us to use bash redirection instead.

Fixes: ffdbecc (ioc: initial commit.)

[ericonr]

I'm not sure I agree this is a fix. Why do you think it is?

[henriquesimoes]

I'm not sure I agree this is a fix. Why do you think it is?

Because I bet no one would use it as busybox <command>. Most users would expect the command to be available in the PATH, and install it from the package registry when the they noticed the command did not exist. If I had noticed this is not the case earlier, I'd have proposed to have symlinks installed from the start.

But I understand that is mostly an extension of the features we provided in previous images, and thus not exactly a bug fix indeed. If you still think it better fits the "new features" section, let me know.

[ericonr]

Commit message changes (please wrap it correctly before committing):

ioc: install busybox symbolic links.

Make all busybox goodies directly available in PATH, to avoid having
to call them from the multicall binary.

This simplifies removal of the wget package from the build image, since
busybox provides its own version, and it's become simpler to use that.
However, the version packaged by Debian does not implement the `-o`
flag, requiring us to use shell redirection instead.

Fixes: 8f02abc2047d5087ce69ed032b4e42006ec56605

I think the main commit being fixed is the one that added the possibility of fetching stuff during build. Having done this for the initial busybox commit would have been helpful, but not critical. What do you think?

And I don't think it makes sense to mention alpine here, since we only use debian for the IOC images.

[henriquesimoes]

I think the main commit being fixed is the one that added the possibility of fetching stuff during build. Having done this for the initial busybox commit would have been helpful, but not critical. What do you think?

I understand you point, but I also meant users podman execing inside containers and attempting to debug something. In such cases, for instance, they would not need to install iputils-ping, but I think they would, because ping would not be in PATH. That's why I referenced to the initial commit instead of 8f02abc .

And I don't think it makes sense to mention alpine here, since we only use debian for the IOC images.

I agree on this one. Referencing it made sense for me at start because I was troubled about why we did not need to install it in alpine as we did for lnls-debian-11-epics-7.