[discuss] How about sections?

[TxGVNN]

After using kubel long time, I see I have facing with Emacs pinky again, although I'm using a split keyboard :(

Because I have to switch to resources, namespace too much. So I thought about how we can display list of resources with namespace on an overview buffer (we can customize the list)

Maybe we can use magit-section to define a section of each resource-with-namespace.

But by using tabulated-list mode, seems we can not handle that, I'm not sure. So I want to open the issue to find a good solution.

[TxGVNN]

In my opinion:

• Still use shell to print output, but we will inject properties at point(or line) each resource to define context, namespaces, resource type and id. So we don't need use the kube-process buffer, we can also resolve https://github.com/abrochard/kubel/pull/57
• Handle any kube-* at point will use info from properties at point/line instead of tabulated-list-get-entry
• Use magit section to define section

[abrochard]

That's an interesting idea. I wonder how kubernetes-el does it?

I picked tabulated-list out of convenience for implementation and also to mirror the terminal experience as my main command was kubectl get pods.

Just so I understand, your workflow involves switching resources a lot so you'd prefer to have all objects at hand from the get go?

[TxGVNN]

That's an interesting idea. I wonder how kubernetes-el does it?

I tried kubernetes-el long time ago, It looks like guix.el interface. But I like kubel because it is KISS. Seems I should try kubernetes-el again to check my workflow is match?

Just so I understand, your workflow involves switching resources a lot so you'd prefer to have all objects at hand from the get go?

Yes, I switch namespaces and resources so much. Have all resource at hand is very cool. Or other solution is having separate buffer mentioned in #57

[abrochard]

Yeah it really sounds like an implementation of #57 would enhance your workflow. You could even whip out a sidebar or a smaller buffer with all the resources you care about.

[d1egoaz]

@TxGVNN I wonder if you can try https://github.com/abrochard/kubel/pull/57

as I'd love to get more test users for this. It seems @abrochard got hit by some bugs that I don't have locally 🤷

I've been using my fork for a while now, working with multiple kubel instances (buffers) with different cluster/namespace/resources with no issues.

I personally use https://github.com/d1egoaz/kubel/tree/test which provides me with a single output process buffer as well. That said, I'll need to find some time to upstream some changes here, but I don't think is going to happen soon, not much time these days, being busy at home with 🧒 🧒

[d1egoaz]

I've tried kubernetes.el and I've found kubel.el is superior as I can navigate custom resources easily and doesn't try to be smart and bring unwanted information.

kubel.el source code is also agnostic to the resources. in kubernetes.el there is special code for each resource, which doesn't scale when you have hundred of custom resources 🤷

[TxGVNN]

Thank you, @d1egoaz, but right now I don't use kubel so much =D. I will try it if I work with k8s again.