rcfox
rcfox on Hacker News
I'm not convinced people acting on muscle memory would remember to unignore the files either. They're going to lose work or have giant "oops, I forgot to commit these files" commits.
This seems like bad advice. I've very rarely committed extra files by accident, but I would 100% forget to unignore files I meant to commit.
If you're doing an initial setup step to gitignore everything, why not just do an initial setup step to gitignore the usual files? Make a template that you copy into all of your repos.
I've noticed that too. Maybe the normal Write tool has to output the entire file and this is an attempt to reduce token usage?
Auto mode feels like a solution to a problem they created. Whitelisting commands used to work fairly well until the agents started generating complex bash commands that it gave up trying to classify.
"Oh you're piping grep into sed, I have no idea what the implications of that are!"
The more complex bash scripts are perhaps better or more efficient, but it feels like it's training us to give up more control to the agent.
I'd prefer to have an auto mode for read-only operations and go back to simpler commands for write operations that I can personally whitelist or manually accept.
Huh, it's wild that they didn't send an email about this. I can't imagine people taking this well, especially as they don't seem to be giving notice about the end of it either.
This is the first I'm hearing about a usage cut. Do you have a link to an announcement for this? Search hasn't shown anything.
It looks like SARIF is JSON, with a formal schema. I'm guessing the JSON they used to output used their own, non-standard schema.
Could be to highlight their incompatibility?
There are communities of people who publicly blog about their eating disorders. I wouldn't be surprised if the laymen's discourse is over-represented in the LLM's training data compared to the scientific papers.
Maybe you need to make shorter PRs?
Yeah, my first thought (admittedly an absurd one) went to something along the lines of:
"I flipped a coin and the LLM called heads. I should have gone with tails..."
The title is misleading. Productivity isn't at 10%, it's at 110%.
> For Linux, just use the emacs that comes with the distro.
Are the major distros shipping packages with tree-sitter support yet?
Your comment heavily implied (to me) scaling databases horizontally. Yes, it's not necessarily "Google scale" either, but it's a ton of extra complexity that I'm happy to avoid. But a Google employee is probably going to approach every public-facing project with the assumption of scaling everything horizontally.
With multiple servers talking to a single database, I'd still prefer to let the database handle generating IDs.
"Distributed systems" doesn't have to mean some fancy, purpose-built thing. Just correlating between two Postgres databases might be a thing you need to do. Or a database and a flat text file.
Do most people? Not everyone is Google.
I worked on creating a card game with a couple of my friends. We kept all of the cards in a Google Sheet, allowing everyone to easily edit or create new cards. Then, I wrote a script to compile the sheet into a card atlas so that it could be consumed by Tabletop Simulator. It worked amazingly well.
The only issue was that I had to run the script myself, since my friends were less technical. I'd probably see if I could setup a workflow in Github Actions to do it for me if I were to do this again.
In Python, every variable is either defined or imported in the file in which it's used, so you always know where to find it. (Assuming you don't do `from foo import *`, which is frowned upon.)
In C++, a variable might be defined in a header or in a parent class somewhere else, and there's no indication of where it came from.
Is this just because LLMs don't have state?
As far as I understand it, as you have a back-and-forth conversation with an LLM, you have to provide the entire history of the conversation plus your new response each time.
With Typescript, I find it pretty eager to just try `(foo as any).bar` when it gets the initial typing wrong. It also likes to redefine types in every file they're used instead of importing.
It will fix those if you catch them, but I haven't been able to figure out a prompt that prevents this in the first place.
These days, I'd say always start with the YouTube tabs.
I also went into that book blind. I was in grade 12 and some organization was offering scholarships to people who wrote an essay about the book. I had a twice-daily 45-minute bus ride to fill, so it seemed like an easy win.
I didn't end up finishing the book.
Ultimately, you need to decide who your target user is. Do you want to cater to the lowest common denominator, or do you want to want to make something power users can customize to fit their workflow?
Neither answer is necessarily wrong, you just need to make a choice.
I welcome being out of touch. The current trends seem ridiculous to me.
There's no practical difference between sending a message via Instagram vs via email. It's all just marketing. You probably can't even get an Instagram account without first getting an email account.