midribbon_action

I appreciate that, I spent time trying to write as clearly as possible, and it hurt to have someone, either through negligence or malice, completely misunderstand me. I still don’t see how the natural language part relates to creating systems without interfaces. An interfaceless system is nonsense, it doesn’t exist, it just betrays a lack of understanding. Even black holes have interfaces. The log files are an interface. The other interface you mentioned was actually a system to aggregate log files, and I just have to point out that systems are different from interfaces. The interface in that case would be the resulting directory of files, or maybe it could be a streaming interface directly into an llm. But anyways the thing that really bugged me was I was asking about the use indeterminate systems to replace solved deterministic problems, and your response was something along the lines of ‘yeah I agree, add more llm!’ and I would be upset if anybody passing through thought that you and I agreed on that.

I’m sorry if my comment came off as rude, I’m having a bad hair day.

Sorry I know you probably don’t want another tip from me, but the post did include the agent directly using the docker daemon, which runs as root typically. Because you didn’t mention running rootless docker or podman, your sudoers file probably allows the agent full access to root instead of preventing it.

I don’t think you understand what an interface is, I don’t think you understood my comment, and my suspicion is you probably can’t or won’t even if I try to help clarify.

Wow amazing, an llm that can generate text!

I’m still curious though if you are going to change your approach after this test.

What answer did you arrive at? Are you planning on ending the test?

I knew dockhand/portainer would do docker updates better, i knew auto updates can be setup via cron for os updates, etc.

Well, that could’ve been mentioned? Why did I have to bring that up? Nothing about the post is self reflective, it is entirely bragging. I get you didn’t write it, and that’s just how llms sound, but you did decide to post it.

Are you working at an ai startup or university? I asked op for their motivations, and comparisons to existing solutions seemed like the least of their concerns, maybe even unconsidered. But I guess it could be a fair answer from your pov if you are trying to test and improve llms. I just hope you’re getting paid for the research.

There is no comparison, I made the comparison myself. In all honesty I feel like they didn’t know about basic docker and linux concepts until my comment.

I guess I’m confused… The built in functionality seems like the easier way to accomplish the same, you seemed to have spent a large amount of time and are proud of this project, and wanted to share it, but also acknowledge that it’s worse than what already exists, and uses more resources idly. Why should anybody else do this?

It seems the main use case is restarting docker containers, why not use the built-in healthcheck feature of docker? The automatic backup and upgrade are also confusing to me, operating systems come with that built in. I just don’t quite understand the point of replacing existing deterministic systems with a natural language interface, I would have trouble believing the logs at face value.