> So, MS builds VSCode - doesn't even fork Atom to do so. Looks identical to it. They built it from scratch. Bigger. Slower.
Someone needs to fix a memory leak here.
Atom was famously slow. Even among people using it and championing it.
VSCode totally wowed people not just because it was faster, but because it was essentially the first «real» Electron-app which proved Electron-apps could have near native performance.
I think a better frame would be «how could the maintainers have responded in a constructive, collaborative way upon learning about the tooling not being compliant with Emacs-standards, in a way which have helped land what was clearly a good faith effort aiming to make Emacs better?»
Outright rejecting the patches was IMO not a pragmatic or constructive choice and will drive the wrong incentives wether you morally approve of it or not.
Did you review the code in question and end up rating it slop, or are you just reflexively calling anything AI-generated slop?
Humans can produce garbage code. As can AI. So therefore the process around the code matters, and it seems clear to me the author has had a reasonable process around the code, as opposed to blindly accepting some 1-shotted output.
This is an impressive effort and offering and I really want to try it out!
Clicking on "download" though, I get this:
> Full edition (121G zipped, 174G unzipped): From Internet Archive
Not to be picky about free stuff offered by others, but I'd be more happy to download a non-zipped torrent, ready for use, where I can contribute BW back to the project itself as a means of gratitude.
While the news is interesting in itself, I found the lack of illustrations disappointing.
When discussing new novel molecular structures, one would think providing a concrete visuals of what they look like more interesting than human-scale photos of materials containing them?
> You are greatly underestimating the current hardware requirements for productive local LLMs.
Fixed that for you. Right now most models produced are based on floating point maths and probabilities, which is "expensive" to do math on.
Microsoft has researched 1-bit LLMs which can run much more efficiently, and on much cheaper hardware[1].
If this research is reproducable and reusable outside their research models, this means the cost of running self-hosted LLMs will be reduced by an order of magnitude once this hits mainstream.
> Very few people with LinkedIn profiles read the social feed.
I read somewhere that in Norway (small sample, yes I know) LinkedIn is supposedly a more popular social network than X/Twitter.
You can have whatever opinion you mean about Elon, X, free speech and whatever. I'm not here to have that discussion.
All that considered, as a Norwegian this had me quite surprised. I don't have the source anymore, but I'd love to dig into it to see what sort of metrics they use to measure this sort of popularity.
Literally nobody I know uses LinkedIn except for business-SPAM.
So I’ve enabled reduced transparency and all the other accessibility settings I can find to remove the terribleness.
The UI is now mono-coloured gray and looks like MacOS back in the days before OS X was a thing - but it’s still better than what Apple “envisioned” with Tahoe.
https://jostein.kjonigsen.net