HN user

ivankelly

62 karma
Posts2
Comments44
View on HN
Knitting bullshit 3 months ago

Yes, I saw. By giving up I meant I skimmed to the end. The images improve the article

Knitting bullshit 3 months ago

I came in indifferent but it doesn’t take much to make me give up on an article linked on hacker news. I use it as bubblegum while waiting for a compile/prompt, intent ally for stuff that can be dropped easily. I saw her disclaimer at the end. My point was that the slop images make a more appealing article than if they were absent

Knitting bullshit 3 months ago

I was a couple of images in before I sussed it. Bullshit images, but pleasing enough to look at. Without the images, it would have either been a big wall of text, which would have put me off reading, though I did give up about 25% of the way through after sussing the images and thus the incoherence in the argument. The images bring something to the article. They were cheap/quick to generate. The increase the potential payoff (more reader) without significantly increasing the cost. Without the images, the payoff(readers) would likely have been lower, below the cost of actually writing the article. Same goes for a history of knitting podcast or that video. Production costs would not be worth it for a very niche viewership.

Not at all. It’s a tool. It can be used well and it can be used badly, the difference often being others things thought in a CS curriculum. But a well trained engineer using the tool will be more productive than that engineer doing everything by hand, so leaving that tool out of the curriculum is doing a disservice to the students

There’s queries and there’s queries. Many queries are effectively undefined bookmarks. You know this exists but haven’t saved it but your know a few key words to get to it. “Rustdoc moka api”. And then there’s the “I’m researching a subject queries”. Google used to be useful for both. But for the later case it send to have gotten worse. Which is irrelevant because LLMs are so much better. So people will use LLMs for these usecases. So Google search basically becomes a phonebook. And that’s easy to emulate. Their calculator is still my go to though

I recall CodeWarrior being the official ide for SymbianOS when I started there. And it sucked, but likely more due to the integration. I think sucky custom rarely working IDEs is what pushed me to full time emacs

At the time of Jurassic Park’s release, the meteor as the cause for the dino-extinction was still a nascent theory and more than a decade out from being cemented as fact.

This assumed me. Like many people who were interested in dinosaurs, the interest didn't last much past by early teens, so the "nobody knows for sure, maybe meteor" reason for their disappearance was the accepted explanation until something triggered me to look a 2 or 3 years ago and see that the science had changed.

Introducing S2 2 years ago

A interesting difference is the ability to have exclusive access to writes on the log (the fencing token). This allows you to use the logs as write ahead logs.

The gripe water that parents in Ireland and the UK used to give to babies used to be a mixture of sugar and alcohol. 3.6%. They took that formulation off the market in the 90s though.

100% agree. I haven't read the raft paper in years, but I remember thinking there's just too much stuff in there. That stuff in important, but if you want people to understand what's happening they internalize the fundamental idea of being able to block other writers by bumping a number. Which is all covered in the single decree paxos section in part time parliment.

Note that in Portugal garbage is usually collected from nearby (unsaturated green colored) lidded trash cans and dumpsters owned by the city government or entities contracted to it

Is this uniform across all of portugal? In spain, it differs by neighbourhood. Where I live, you drop landfill rubbish beside front door between 20:00 & 22:00 and it gets picked up. 100m away in another neighbourhood they have underground depots that you drop it into. In the newer parts of the city they do have containers on the corner of each block.

Since they're already doing offload to S3 for older log messages, it seems like they could do this from the get-go and save a bunch of storage and interzone transfer costs in RDS. Depends on the log message sizes though of course.

The missing dimension here is the percentage of time the program is running on the core. Even if the benchmark is using 3 seconds to run 11billion instructions, that doesn't tell you if the core is still idle 90% of the time because it's blocking on I/O. CPU bound work should pin the core. This is especially true on server side stuff, because if you are not maximizing the time on core, you are paying for CPU that is not being used.