Happy "None of Us Will Be Shoveling This Snow" Day for all who celebrate, which is all of us.
Happy "None of Us Will Be Shoveling This Snow" Day for all who celebrate, which is all of us.
Meetup has been getting worse for awhile. Someone should make another one and start the cycle anew.
A hot take: our perceptions are skewed by the industrialization of clothing. A world where making clothing by hand is the default would normalize the mistakes more
So much this. I know the standard advice is "nobody will care or see", but these hobbies are all about the act of creation - taking raw material and breathing life into something via human effort. So it makes sense that you'd want it to be something of high quality
I'm at this weird middle point with sewing, where my ability to perceive issues has grown beyond my ability to actually not make those mistakes in the first place. Currently trying to get past that so that I can enjoy the process and not care so much
I'm realizing that the trick to enjoying any hobby is just an enormous amount of hubris, combined with radical acceptance of flaws in the result
I've been scraping the metro transit GTFS feed for about 36 hours and it's kind of lovely to see the activity ebb and flow over the course of the day
I'm in this photo
And before we go saying "well the database is really impressive"
- no evidence that it's being used anywhere critical
- and after glancing at the code, I would not.
I don't wanna argue too hard against the whole factory advice, since I think there's good ideas in there.
But if I'm gonna try and emulate their practices, I would like more confidence that it'll actually work
I am low-key unimpressed by the StrongDM AI thing. I would expect some extremely impressive output given the hype, but so far I see... Oauth for agents, and a custom DB for convo history?
(Am I missing something obvious?)
But yeah I'm in the same boat as you: I've been creating my own thing from a bunch of pieces, and I keep waiting for _something_ to put all the pieces together in one box
I've ended up on zellij/tmux for this, with helix/vim as an editor.
AFAICT nobody has solved the "place to comment on the diff" part in a good way yet, I just push everything to GitHub and wrote a script to fetch pr comments and get markdown back.
Counterpoint: hopefully we get to the point where our tokens/sec is fast enough to not need this. Why bother parallelizing when each task is done quickly
Gastown (supposedly) solves the "many agents integrating with each other" problem and I think that's a pretty huge deal - I would really love it if GitHub let me throw Claude at solving merge conflicts automatically
I mean you definitely get to the point where one coding agent just doesn't go fast enough. I was cooking up a side project this morning & was waiting on Claude to spin on a problem so I could give it the next one
Yeah big ++ on that, I will admit to not having used it, but it feels like the purpose of gastown is to make people understand what's possible instead of being the thing
Yeah I see the utility but I think I will wait the 6 months for someone to figure out how to deploy one safely.
The real worry here isn't the security of the harness, but the prompt injection attacks you're opening up by giving an LLM privileged access to your life, and an outbound Internet connection
People hear this kind of rhetoric and they remember it for a very long time. Dismissing someone's work as being less difficult or technically challenging than your own creates strife
Maybe I haven't worked in ops long enough but this feels deeply embarrassing as someone who has for a bit.
Hard problems can be found everywhere! Ops is hard! Product dev is hard!
Trying to fight about which one is harder or more challenging is just putting unnecessary divisions between teams
At some point the property tests I can come up with just feel like some extra ceremony around example tests, but then I lose nice properties like being able to easily debug!
Part of this is that property testing feels magical - the examples are always for things like "assert that is sorted(sort(x))"
Which is nice _in theory_, but the docs don't really have much in the way of more complicated examples.
I use them for eating chips and most snacks, but my chopsticks skillz aren't sufficient for popcorn. Also taking a small handful of kernels and eating them at once is half the fun of popcorn, for me.
For once in my life it would be cool if the linux OOMKiller made the correct choice in a timely manner
Itβs that time of the year where Home Alone is going to be on tv a bunch. As relatives go βthey should have had a battery powered alarmβ and make other judgments on the familyβs lack of responsibility in forgetting their kid, get a better source to learn from what happened.
ferd.ca/home-alone-a...
Yeah you should be teaching them helix
nix-direnv registers gc roots, so by default the things you downloaded for old side projects doesn't get cleaned up
FYI if you're like me and use direnv + nix for projects, do yourself a favor and clear out the `.direnv` folder in your old projects and run `nix-collect-garbage` - you might find out where all your disk space has been going :)
Growing up there def gives you a biased perspective, but if you ignore a very healthy amount of political/cultural issues, it's a fun place to spend 2-3 days.