I recently watched the 3.5-hour workshop Mastering the curl command line by Daniel Stenberg, the author of curl
I commend the author for watching and subsequently summarizing this into a blog article, but in the age of LLMs like ChatGPT I cannot be arsed anymore to memorize any of the arcane command line arguments of curl, ffmpeg and the like.
You don’t have to memorize. Next time, you can just recall the commands by looking at manpages or this blog post again. You will already retain most concepts.
ChatGPT is cool, but it is wrong often enough that makes hard to trust. I don’t want to be running the wrong command and suffering its consequences. I only resort to chatGPT when docs and web search do not give me an answer quickly. Even then, I try to verify chatGPT with docs before going forward.
I mean, we’re talking curl here, I don’t think a lot of suffering of consequences is happening. And man pages are often also not a great resource, throw everything at you, often don’t contain examples.
If I’m building an app that integrates curl or libcurl, oh yes I’m reading the doc. If I just need curl to do something quickly, the LLM output is perfectly fine.
I think you’re right, it is harmless for the most part. But one time I unintentionally overwrote a file using curl. It is definitely my own stupidity here though.
Who said anything about trusting blindly? We’re talking curl here, not dd or fdisk. And in my experience the LLM gets it right most of the time, and even if it doesn’t, it’s often still faster to just test the outputted command and feed back the error than crawling through the manages.
With CLI, all you need for most use cases is “command --help” or “tldr command” or the beginning of “man command”. Been using curl for ages with just that. This blog post is only if you want something in depth to master it, not a requirement to use it.
A GUI is all about figuring things out and it is often pretty clunky compared to a cli, imo.
See, I need a two hour tutorial to use the GUI because everything’s scattered all over the place. If I want to know the options for a CLI tool I just read the man page.
Until the time comes where the GUI just doesn’t have the specific option for what you want to do, and then usually you can string commands together into scripts to do it. That’s where its power lies. Every command/utility on a Unix/Linux system is meant to be a tiny building block to use to accomplish a more complex task.
You can’t possibly program an option for every contingency into a static GUI, but you can hand the user a toolbox of command line utilities and then they can accomplish nearly anything.
I commend the author for watching and subsequently summarizing this into a blog article, but in the age of LLMs like ChatGPT I cannot be arsed anymore to memorize any of the arcane command line arguments of curl, ffmpeg and the like.
You don’t have to memorize. Next time, you can just recall the commands by looking at manpages or this blog post again. You will already retain most concepts.
ChatGPT is cool, but it is wrong often enough that makes hard to trust. I don’t want to be running the wrong command and suffering its consequences. I only resort to chatGPT when docs and web search do not give me an answer quickly. Even then, I try to verify chatGPT with docs before going forward.
I mean, we’re talking curl here, I don’t think a lot of suffering of consequences is happening. And man pages are often also not a great resource, throw everything at you, often don’t contain examples. If I’m building an app that integrates curl or libcurl, oh yes I’m reading the doc. If I just need curl to do something quickly, the LLM output is perfectly fine.
I think you’re right, it is harmless for the most part. But one time I unintentionally overwrote a file using curl. It is definitely my own stupidity here though.
I’d be scared to perform POST/PUT with LLM-generated commands. For immutable calls I agree though
you should beware trusting blindly any LLM output, they are often wrong
Who said anything about trusting blindly? We’re talking curl here, not dd or fdisk. And in my experience the LLM gets it right most of the time, and even if it doesn’t, it’s often still faster to just test the outputted command and feed back the error than crawling through the manages.
deleted by creator
With CLI, all you need for most use cases is “command --help” or “tldr command” or the beginning of “man command”. Been using curl for ages with just that. This blog post is only if you want something in depth to master it, not a requirement to use it.
A GUI is all about figuring things out and it is often pretty clunky compared to a cli, imo.
See, I need a two hour tutorial to use the GUI because everything’s scattered all over the place. If I want to know the options for a CLI tool I just read the man page.
Until the time comes where the GUI just doesn’t have the specific option for what you want to do, and then usually you can string commands together into scripts to do it. That’s where its power lies. Every command/utility on a Unix/Linux system is meant to be a tiny building block to use to accomplish a more complex task.
You can’t possibly program an option for every contingency into a static GUI, but you can hand the user a toolbox of command line utilities and then they can accomplish nearly anything.