Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
cube2222
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
32 ms
·
421.
▲
by
cube2222
3y ago
This demo is frankly one of the most mind-blowing things I’ve seen so far in the recent generative-AI boom. Could be quite cool to use in games, esp. competitive online ones, for a more dynamic announcer.
422.
▲
by
cube2222
3y ago
I believe the author acknowledges that a bit later. > The resulting noise is gaussian, not uniform. […] This is why you need a specialized algorithm. (The algorithm is afterwards described)
423.
▲
by
cube2222
3y ago
> but I don't think they've got anyone working on the language server either. Not yet, but it's on our radar. We're focusing on the stable release + registry for now (eta mid-December), but we're planning to work
424.
▲
by
cube2222
3y ago
I wonder how much this is GPT 4 actually inferring the details about what's on the screen vs GPT 4 recognizing the "shape" of Blender, getting the task (which is simple and has most likely a myriad tutorials online), and just
425.
▲
by
cube2222
3y ago
Oh, yeah, that does make sense, and I suppose it might also be a better approach if the objects were large (especially if you can avoid reading some parts). In that case prefetching could be wasteful (or result in OOMs). I didn’t think of t
426.
▲
by
cube2222
3y ago
I might’ve been unclear, so to clarify: The overhead of fetching from S3 via a naive Go implementation (goroutine per object) to disk and then running duckdb on that was lower than using duckdb end-to-end. I was measuring the S3 overhead in
427.
▲
by
cube2222
3y ago
It would be interesting to see some more benchmarks of e.g. querying multiple files from S3, and how that evolved across versions. When I checked at 0.7.1, when working with ~90 S3 parquet objects (x0000 rows each, so not too many) it was 2
428.
▲
by
cube2222
3y ago
> and in retrospect wish I had gone with Nebula Could you expand why (happy Tailscale user here, asking mostly out of curiosity)?
429.
▲
by
cube2222
3y ago
Tip for those wanting to play it: change resolution scaling from dynamic to constant. I have a 3080 and it basically moves it from "unplayable 10fps in the main menu" to "works just fine, no issues in game" with medium-h
430.
▲
by
cube2222
3y ago
Spacelift/OpenTofu | Remote | Europe/Americas | Full-time | Open Source Software Engineer We're a VC-funded startup building an automation platform for Infrastructure-as-Code, adding a Policy-as-Code layer above it, in order
431.
▲
by
cube2222
3y ago
It doesn’t seem like an actual example of this is linked anywhere in the article, so here you go[0]. Generally the „products” share a prefix with one of the parties. For example, KO (koalicja obywatelska) - kolendra (cilantro). [0]: https:
432.
▲
by
cube2222
3y ago
I recommend checking out Talon Voice[0] if you’re experiencing any wrist (or similar pain). It’s a very intuitive and customizable voice coding tool, and even if you don’t need it at all times, you can use it to lower the strain on your han
433.
▲
Zig Interfaces
(openmymind.net)
10 points
by
cube2222
3y ago
|
0 comments
434.
▲
by
cube2222
3y ago
Spacelift/OpenTofu | Remote | Europe/Americas | Full-time | Open Source Software Engineer We're a VC-funded startup building an automation platform for Infrastructure-as-Code, adding a Policy-as-Code layer above it, in order
435.
▲
by
cube2222
3y ago
I believe that is actually a recently-added feature of the Hypervisor.Framework, not sure if parallels uses it under the hood. It's also been in Docker Desktop for a few months now, thanks to that.
436.
▲
by
cube2222
3y ago
I used to have this problem, thinking I just can't get accurate in FPS's. Then I read that your mouse sensitivity should generally be around 30-60 cm movement to do a 360 rotation. Mine was more in the area of 5-10 cm for a single
437.
▲
by
cube2222
3y ago
> Maintainers act like they own the thing and everyone else (read: the users, as opposed to the developers) can just suck it. > project doesn't actually care about the users > Community definitely isn't the goal Hey, Open
438.
▲
Why should I care about OpenTofu?
(opentofu.org)
3 points
by
cube2222
3y ago
|
0 comments
439.
▲
by
cube2222
3y ago
Thanks for the kind words! The initiative will have a steering committee composed of individuals from the main backing and involved companies and projects. As interim technical lead I'm mostly responsible for the technical side of thin
440.
▲
by
cube2222
3y ago
Thanks for noticing, we need to update the org description.
441.
▲
by
cube2222
3y ago
Yep!
442.
▲
by
cube2222
3y ago
Hey! Interim tech lead of OpenTofu here. I'd like to add that as of today (announced just now at OSS Bilbao) we're officially part of the Linux Foundation[0]! Hope you like the new name (it basically won the votes anywhere it was
443.
▲
by
cube2222
3y ago
This is by the way kind of how remote access with apple home works. The Apple TV serves as a local gateway relaying all the commands to your local IoT devices. On a side note, tailscale is lovely. I have nothing but good things to say about
444.
▲
by
cube2222
3y ago
> and the fact that things are breaking the test suite left, and right Branch protection doesn't allow merging without a passing test-suite. > The squash merge is not going to solve the lack of proper commit messages Could you ex
445.
▲
by
cube2222
3y ago
Even though you're being downvoted I do agree that this should've been squashed (I don't see any other problems here, if that's not it). I've made sure via repo config that only squash commits are enabled from now o
446.
▲
by
cube2222
3y ago
I'm not sure which docs you mean specifically, but most of the docs are fairly unchanged (other than trademark stuff) from the original repo, so if any of the docs improved, then to give credit where credit's due, the kudos should
447.
▲
by
cube2222
3y ago
Hey! > The first is, please provide a standalone registry package for both modules and providers - the only I'm aware of is Artifactory, and I don't really feel like running another big repository software in a Nexus shop. Coul
448.
▲
by
cube2222
3y ago
Hey! As many of you asked for, we've finally made the repository public and will continue developing it in public from now on. This took us some time, but you can now read our announcement for more details[0]. Thanks everybody for the
449.
▲
OpenTF repository is now public
(github.com)
510 points
by
cube2222
3y ago
|
177 comments
450.
▲
by
cube2222
3y ago
Spacelift | Remote | Europe | Full-time | Senior Software Engineer We're a VC-funded startup building an automation platform for Infrastructure-as-Code, adding a Policy-as-Code layer above it, in order to make IaC usable in bigger comp
More ›