HN user
That's only true if the engineers are not allowed to copy/steal from existing designs. There are plenty that are better than GIMP (e.g. Photoshop, Krita, ...). If nothing else, make it easy to build a layer on top so that Photoshop can be replicated nearly exactly.
They didn't say "reality must be wrong". They said that their initial hypothesis (that significant changes would be observed after 21 days) is probably wrong, so they implicitly proposed a second hypothesis (that significant changes occur after e.g. a few months).
None of this is remotely contemptible.
Pretend you're an immortal alien conducting a study with the hypothesis, "humans are mortal". You observe that your subjects do not die after 21 days. Do you conclude that humans are immortal? (I hope not. It's much better to conclude that humans don't usually die after 21 days in this particular instance of extraterrestrial captivity.)
The author of that post wrote "difftastic", which is "a structural diff that understands syntax" using treesitter.
Will making (non-)computer viruses illegal sanitize the world of them?
A few years ago, the main advantage was merely that "Sidebery is far less buggy and much more performant than Tree Style Tab".
Nowadays, the gap has... widened. The little details matter, but there's big feature differences too: panels, saving/restore to JSON, better appearance, tons of options for tweaking (but with sane defaults), ...
A very interesting property:
Knights have odd-even parity, i.e., they attack only opposite colored squares.
Pop quiz: How many knights can you place on a board so that they don't attack each other?
Personally, I would prefer the conventional Latin Modern math font instead of Palatino math.
Latin Modern is used by:
- Wikipedia. - Math.StackExchange. - Nearly all papers, including the ones hosted on arxiv in PDF format. - Nearly any math videos, slides/presentations, notes. - Almost everything, really.
Palatino just looks weird.
Also, I imagine that authors might do math formatting hacks that were only tested on Latin Modern, and might end up breaking on Palatino.
TL;DR:
Palatino :(
Latin Modern :)
How would an AGI gain access to an unlimited credit card that immediately gives it remote access to all GPUs in the world?
It's also quite silly that society often credits one guy at the top who supposedly has "incredible vision" and yet would likely fail at explaining even the most basic technical details. And if such a person must be credited, why not the CTO, chief engineers, or principal scientists, who are at least closer to what actually drive the technical innovations than the CEO?
In reality, it's actually the 1000s of actual engineers that deserve most of the credit, and yet are never mentioned. Society never learns about the one engineer (or team) that solves a problem that others have been stuck on for some time. The aggregate contributions of such innovators are a far more significant driving force behind progress.
Why do we never hear of the many? It's probably because it's just easier to focus on a single personality who can be marketed as an "unconventional genius" or some such nonsense.
See "concert earplugs" that try to reduce the volume with minimal distortion.
64k tokens ought to be enough for anybody.
`jj split -i` gives:
error: unexpected argument '-i' found
Actually, maybe I'm just a complete git, but I couldn't figure out how to `git reset HEAD~` my accidental commits, `git rebase -i HEAD~6`, format `jj log` more like `git log --color --oneline --graph --full-history` (which shows one-line-per-commit), `git checkout -p` (and obviously, `git add -p`), `git show HEAD~`, refer to N-th parents, e.g. `master~5`, and a bunch of other things...
It also feels a bit weird that new files are automatically committed without being manually approved, but I suppose this might theoretically help with some of git's annoyances.
Neat tricks. Beyond BufWriter (which I'm already using) and multthreading, I'm guessing there's not much to be done to improve my "frece" (a simple CLI frecency-indexed database) tool's performance without making it overly complicated. https://github.com/YodaEmbedding/frece/blob/master/src/main....
Technically impossible since queens cannot materialize out of thin air, except in crazyhouse.
Also Rg1+ is a rather pretty attraction/clearance sacrifice/double check/mate-in-2.
Nearly no one inhabits Greenland and Antarctica, which might help explain why the only visible relative increase in Arctic/Antarctic nationalism comes from enthusiastic polar bears and mildly racist penguins.
I'm not sure there could even exist a particularly satisfying source for this. One of the issues with "soft" sciences is that it can be quite tricky to measure any effects, much less design a viable study that demonstrates causation.
I suppose a social scientist (i.e., not me) could support this claim using ideas from psychology or finding related studies. But I doubt anything will ever be particularly convincing unless we lived in a universe where people told the objective truth and a mandatory survey was asked with the explicit question, "Have the distortions induced by map projections influenced your beliefs regarding people in other countries?"
The issue isn't possible bias on the part of educated cartographers and differential geometers. It's the bias of the majority, i.e. everyone else, that matters. A vast majority of people in the "Western world" perceive their countries as being bigger than they are. Thus from a primal, tribal perspective this further inflates their perception of "their peoples" being more important than the peoples of other areas.
From a non-expert's standpoint, Vulkan feels quite unusable and complex.
Ideally, arrays should start at 0.5 -- that is the best compromise.
Or maybe start with 0.25, since starting with 0 is three times better than starting with 1.
The main thing I miss in these vi-like browsers is a tab tree sidebar like Sidebery.
I guess JIT type specialization / "monomorphization" might be one way to do it, but I expect that it would be bottlenecked by all the surrounding CPython interpreter stuff.
/amateur_two_cents
Replication does not help in managing completely. That's why we use abstractions, even with the problems they have.
For example, it only took 20 years (!!) to stop Ctrl+Q from quitting Firefox on Linux. :)
IIRC, a couple of patches did get submitted, but never accepted for unknown reasons.
This happens on Linux too. I was wondering if the weird CPU-hogging flickering was a bug in my compositor (picom) or window manager (i3) or browser (Firefox). Turns out to be a "feature".
The "average color" (or whatever it is) could have been pre-computed server-side rather than tiring out the poor innocent client CPUs.
Perhaps the ResNet paper, which has the most citations of all time.
If IP stops being a thing, we can always invent new laws to replace it such as "impersonation".
With Sidebery (modern alternative to TST), the tab tree is automatically backed up every $time_interval.
For a given repository, is it possible to have multiple worktrees for different "branches"? (Just like git worktree.)
Python 3.11 made some updates to the `import antigravity` feature.
New quote just dropped:
Premature complexity is the root of all evil.