← Feed

Posts

• Post #4378694

  so I like testing a lot I'm thinking a "mc/dc" (hipp's slqlite talk - testing) but in ac/dc style FOR THOSE ABOUT TO TEST WE SALUTE YOU ;)

• Post #4355961

  "prevent cognitive debt by manually retyping LLM-generated code" is the most absurd headline I've seen this year

• Post #4354489

  what LLM usage drives home for me is giving a flying fuck about "but you have to do/use X, it's best practice/nobody knows Z/there a framework for it" and really go back to doing programming with programming like you know actually doing "the code" however I (we) think is right and good with whatever language I (we) think is nice and a pleasure to work with because it's my (our) freaking shit and I (we) can do whatever I (we) want prior art would be janestreet/tigerbe...

• Post #4353264

  teleprompter is available as a job description maybe senior cyber teleprompter

• Post #4241761

  Just so we're clear - before you think that the big re-hiring is about "AI doesn't work" - judging by job ads and offers I'm getting - it's "throwing more AI engineers onto AI" your old job won't come back in the next 5 years at least - if ever - and even if it's backfilled, it'll be "AI DevOps" or "AI Release Engineering" or "AI Engineering". If there's big hiring, it'll also be big re-shuffling.

• Post #4197358

  On the subject of looking for a new job: I'm VERY put off by fedihire posts which go like "I want to do X (X usually being Rust) but I would EVEN do Y which I haven't done in 20 years ha ha" (Y usually being PHP or Java). Are people actually still thinking this is a cool and acceptable attitude to belittle PHP (or Java or MySQL or..)?? This makes me so SO angry because we ARE DOING IT TO EACH OTHER. Like the software industry as a whole doesn't have enough problems alr...

• Post #4022732

  I've learned over the last 2 weeks during my quest to teach people to use LLMs to not use LLMs ;) that a lot of people just don't know how they didn't know what a thing like shell one liners exists (and that there's a ton of websites about it) they didn't know that they can give it a name (alias) they didn't know that you can treat your computer like a database (osquery) they aren't even used to old school asking "I want to do this thing and then this and th...

• Post #3889345

  spectacular AWS billing incident where people are seeing BILLION dollar high estimates omg https://www.reddit.com/r/aws/comments/1uyuaw7/help_my_bill_skyrocketed_from_around_5_cents_per/ edit: SEVEN TRILLION

• Post #3881523

  I'm VERY curious (if you can tell me) if and how your company does local-and-smol models - did folks just get a chunk of openrouter (or equivalent) to try all the models or did folks just get a handful of GPUs and llama'd it all into place or... I'm particular interested in regular/small/average companies and their decision making process around this IF you decided to actually use LLMs

• Post #3880941

  I have a feeling that it won't be "more Claude" that's deciding over whether or not LLMs are "here to stay" but DeepSeek, Kimi, GLM, Mistral etc etc Any regular, average company who actually WANTS to have LLMs is IMHO better off with the other models - or even chooses to go local-&-mini models, purpose-made which don't cost you an arm and a leg of qualified developer. Set aside the choice with whom to do business or run it yourself entirely. It seems Germ...

• Post #3855800

  So I've had two tiny mini micro successes with teaching to use LLMs much less in the sense of "only once instead of repeatedly" basically by teaching a mix of "automation - so useful!" and Alan Keys idea of writing more DSLs per project (which easily translates into Shell or Python scripts) I couldn't convince any of the juniors however to really dig into shell scripting and spent the effort on learning it for real. To them, it's like argueing that learning Lat...

• Post #3846168

  isn't getting away from rms reason enough for human-emacs usage why wait for an AI policy

• Post #3843884

  your "good old times when the internet was awesome and great code was written" are my "ultra-sexist assholes on the internet and being asked at great code events who my boyfriend is" times. maybe a little less reminiscing and a bit more reflection

• Post #3757449

  Thinking about harnesses (again): any language which has smooth and cheap spawning/forking/co-whatevering is awesomely suited for all the sub-agent concepts I know totally obvious but it made me realize (again) the power of shell'n'tools and how fast a harness in let's say erlang could actually scaffold something pretty big etc etc I also realize (again) how many things could be formalized in the harness without ever touching an LLM Really sharpens "timeless" programmin...

• Post #3612468

  looking at the #cobol first person shooter and it's surprisingly readable https://github.com/icitry/FPS.cob/blob/main/fps.cob

• Post #3487734

  me too donnie, me too... #donchonk #caturday

• Post #3353355

  Here are two articles I think are reading-worthy if you are interested in some ideas where LLM things might be going: https://vickiboykis.com/2026/06/15/running-local-models-is-good-now/ https://blog.alexellis.io/local-ai-is-not-opus/ This somewhat aligns with the general vibe I'm getting from reading HN threads which are sounding distinctively "sobering up". Haven't seen a hype thread in weeks.

• Post #2180205

  As a Linux user for 30 years now, I'm available for somewhat opinionated advice if (in particular women) you have questions. emphasis on polical choices to the best of my best knowledge. advice strongly "it has to work for YOU and YOU have to like it" flavoured. AMA.

• Post #2128652

  well my name isn't claudia but I am as capable as a large language model and have attracted men and women with my linguistic capabilities and made them fall in love also, my availability is still better.

• Post #2128651

  quote of the day "brevity is the language of money"

• Post #2128650

  I find cleaning up incredibly liberating.

• Post #2128649

  two weeks ago I got the vibe that maybe the AI stuff is now declining but I think it's going to go into overdrive now...

• Post #2128648

  tested "overnight oats" for porridge. didn't save anything, still needed some minutes of cooking. one more dish dirty. 👎

• Post #2114397

  in terms of advanced pun and political meme and overall funny my award of the year goes to @klasse_methode I laugh every time it scrolls by in my timeline 😂 (socialist red-black design and phrasing combined with technology pun - works particularily well in german - the literal translation "class and method - IT collective" isn't quite doing it so well.. :))

• Post #1889630

  I didn't look at news for like 3 hours and actually did some work and what happens? UAE left OPEC, nvidia says AI is more expensive than human workers, oh wait github going to shit is not really news...

• Post #1889629

  this day is utter shit so far so after a break and a second coffee and prescription-level consumption of cookies I'll try a restart. I really started this day completely wrong and computers are stupid anyways.

• Post #1889628

  Another week where I have four nines and neither anthropic nor github do... ;)

• Post #1889627

  who needs recognition of their work... 💔

• Post #1889626

  on a rare occasion of actual irony somebody hacked an llm harness in 400 horrible lines of awk and called it "slop cannon" and I think that is as beautiful as "stochastic labubu" ;)

• Post #1889625

  my top creativity tool: comfy chair or park bench, a seat on a train or the reeling of a ferry and staring into the void. (landscape or seascape voids preferred) alternatively knitting or ironing I notice over and over that I'm a slow person, I need time to think and let my mind wander and mull things over