Live data from Hacker News

Command line functions around OpenAI

kadekillary.work

241–250 of 316 posts

Re: Command line functions around OpenAI

#241
post #227

Wow. 210+ comments and maaybe 10 of those actually talk about the article! (yes, I've read them all because I can't believe almost everyone missed the point) The point of the article is actually neat: a small shell (Fish) script to send query to ChatGPT or DALL-E and get response back in the terminal. As a terminal user myself, this is really nice! (I rewrote it for myself in bash which I use, but used the curl comma…

[dead]

Re: Command line functions around OpenAI

#242
+1 to all the comments about deciding what problems to tackle.

My nuanced 2c on it is that every line of code you write is a liability. Part of what revolutionized pricing mechanisms was FOSS and SaaS, which are 2 strategies of amortizing the cost of development across many many people. Every line you write for your company is a line with amortization across exactly 1. So first consider the value/cost tradeoff of FOSS and SaaS to see if you can compete. 1000x might be simply glueing together a few free/cheap systems which would have cost your company quarters. Putting in nice interfaces/adapters means you can easily cut them out if they become a regrettable decision.

Re: Command line functions around OpenAI

#243
post #10

Being efficient at writing code can barely make someone a 5x engineer. A team of people who write code effectively is easy to do. Hyperproductivity in software is all about deciding what problems to tackle. Richard Hipp isn't worried about chipotle restaurant orders in golang, he's worried about how to store data reliably. That isn't a coding puzzle that ChatGTP is likely to help with. Either ChatGTP can do the whole…

I briefly worked with a software engineer who was making about a million dollars a year. I mentioned that an approach would be a lot of work. His response: “It’s just typing. Typey typey type (makes fingers on keyboard gesture).” Absolute lightbulb moment for me. To become a senior engineer is to realize that the code writing part is inconsequential as long as there are no technical unknowns in the work. It’s freed m…

I often suffer from impersonator syndrome, my freelance rates should be North if 70 per hour but I keep getting stuck at 40...

I grew up poor though and teaching myself coding moved be to earn more then I thought I'd ever earn without a college degree.

I knew about for loops almost a decade before I decided web dev would be my career, not just my hobby.

it's insane that you could get hired and not know how they work.

linked lists, sure. I still don't fully understand those lol, I was going for an associate's and that was in my c++ class but I've mostly only touched PHP, python, ruby, JavaScript... playing around with go and rust some as well, but I've never had to really think about linked lists, loops on the other hand is a daily thing.

hell, not knowing loops is like not knowing how to increment, like seriously, his to add 1 to an int.

Smh....

Re: Command line functions around OpenAI

#244
post #157

One can also decide to just be a https://1x.engineer

10x being real or not aside, if there's a 1x-type they certainly don't do many of the things in this list in my experience and certainly not:

>Writes code that others can read.

>Reads the Docs.

>Updates the Docs.

In my experience, half of the differentiation between above average and solidly "1x" or middling devs is reading documentation and obtaining a deeper understanding of the tools, language, platform they're developing on top of.

Re: Command line functions around OpenAI

#245

Earlier quoted context omitted.

I have to ask: how do you work as an eng who pulls in $1m/yr?

High level IC at big tech can get you around that neighborhood

At semiconductor companies we have plenty of IC's doing IC design. The initialism has multiple definitions.

Re: Command line functions around OpenAI

#246
post #41

Earlier quoted context omitted.

Not yet; I spend 60% my day reviewing PRs. I don't have to if chat GPT gets integrated as a PR reviewer..

ChatGPT will turn your day into spending 90% of your time reviewing code...

That's actually how I spent a lot of my time. GPT drafts code of mixed quality, and I spend a lot of time reviewing it.

What's interesting is that GPT isn't trained to produce high-quality code. It's trained as an autocompletion tool. I'd be curious how smart such a system could be if we knew how to engineer it for quality.

Re: Command line functions around OpenAI

#247

Earlier quoted context omitted.

I have to ask: how do you work as an eng who pulls in $1m/yr?

(throwaway I guess) I make over $1m as a SWE. “Senior Staff” or “Level 7” role at a big tech company everyone knows. I mostly write code! I am way faster than most others here, and I have encouraged and mentored the people on my team to be similarly efficient. I also help decide technical direction but it’s far from the “architecture astronaut” stereotype that writes no code. It’s an infra team so no significant infl…

> mentored the people on my team to be similarly efficient

What were the main changes you introduced/points you mentored them on to get their productivity to also be similarly efficient?

Re: Command line functions around OpenAI

#248
post #229

Earlier quoted context omitted.

I think code-generation is a red herring. I'm more interested in things like: - Explanation/research ("how does this work?") - Code analysis ("tell me if you think you see any bugs, refactoring suggestions, etc in this sprawling legacy codebase") Things that feed into the developer's thought process instead of crudely trying to execute on what it wants

Part of why I don't use ChatGPT very much for work is that I don't want to feed significant amounts of proprietary code into it. Could be the one thing that actually gets me in trouble at work, seems risky regardless. How is it you're comfortable with doing so? (Not asking in a judgmental way, just curious. I would like to have a LLM assistant that understood my whole codebase, because I'm stumped on a bug today.)

I'm not doing it right now, I'm more imagining a near-term product designed for this (maybe even with the option to self-host). Current LLMs probably couldn't hold enough context to analyze a whole codebase anyway, just one file at a time (which could still be useful, but)

Re: Command line functions around OpenAI

#249

Earlier quoted context omitted.

I think code-generation is a red herring. I'm more interested in things like: - Explanation/research ("how does this work?") - Code analysis ("tell me if you think you see any bugs, refactoring suggestions, etc in this sprawling legacy codebase") Things that feed into the developer's thought process instead of crudely trying to execute on what it wants

Unfortunately, LLMs are still prone to make facts up, and very persuasively. In fact, most of non-trivial topics I tried have required double-/triple- checking, so it's sometimes not really productive to use chatGPT. You are correct that I made an error in my previous response. I apologize for the confusion I may have caused in my previous response I appreciate you bringing this to my attention. I apologize, thank yo…

Which is exactly why this use-case would be better. "Give me suggestions/ideas, and then I'll take them under consideration"

Re: Command line functions around OpenAI

#250
post #203

Earlier quoted context omitted.

Add: I only want the CSV, don’t describe what you are doing.

Heh nope, I still get the whole "Sure, here's an example of a CSV file with 10 rows of sample Chipotle restaurant order data:" I wonder if the web UI has some additional "explain everything you do" prompt injection stuff going on.

My favorite trick is to ask for JSON, and tack this onto my prompt: "Return JSON, beginning with {"
Post reply on HN