@karolherbst@chaos.social cover
@karolherbst@chaos.social avatar

karolherbst

@karolherbst@chaos.social

Linux Graphics Developer
Freedesktop Code of Conduct Enforcement team member

Mostly working on Rusticl and Nouveau
Implemented OpenCL in Rust for fun

🏳️‍🌈🏳️‍⚧️ are welcomed.

Nazis, $hitcoin cultists, Right-Libertarians, Longterminists, Tankies, techbros and other fascists not welcomed. This is a shithead free zone.

Private account, please direct all business inquiries to: https://twitter.com/karolherbst

#mesa
#nouveau
#nvk
#opencl
#rust #rustlang
#rusticl
#coc

This profile is from a federated server and may be incomplete. View on remote instance

Len0w0ThinkBad , to Random
@Len0w0ThinkBad@labyrinth.zone avatar

It really bugs me how so many people act as if there's a magic "click here to implement [feature]" button, and completely discount the fact that Free Software like Linux is largely developed by unpaid volunteers.

Yeah, sure, they shouldn't need money and resources to implement accessibility features, they are super simple! Every other OS* has them!

*(Windows and Mac, both developed by small mom and pop shops)

karolherbst ,
@karolherbst@chaos.social avatar

@Len0w0ThinkBad there is an interesting aspect to this and that's getting FOSS/Linux used by governments (as in, paying local IT firms instead of big corpo). However they usually have high requirements on accessibility, which ultimately ends with a chicken-egg problem, because only big companies (e.g. Red Hat) are able to promise on those aspects.

So there is some chicken-egg problem right here, but I think that's solvable if governments would be willing to invest big in Linux accessibility.

Lyude , to Random
@Lyude@queer.party avatar

We need to start teaching classes on AI not because AI is the future, but because as far as I can tell the majority of people are literally incapable of understanding what an AI actually does.
It feels like when people see ChatGPT mis-answer something they think "ah must just be a bug, it'll get fixed".
No. That's not a bug. It gives bogus answers because it doesn't do anything but generate a sentence that looks similar to what an answer to your question could look like

karolherbst ,
@karolherbst@chaos.social avatar

@Lyude my biggest fear is that AI grants some kind of "human" rights and then companies will treat is as such and make the AI responsible for everything and the AI goes to jail "because it has feelings you know, just like humans" and ....

karolherbst ,
@karolherbst@chaos.social avatar

@Lyude I think my best analogy on what's happening right now is, that it's all a big magic trick and people just don't know the trick, but also just don't want to know, because being amazed is more fun than knowing the truth.

sima , to Random
@sima@chaos.social avatar

executive dysfunction is doing something in like 1h that you've put off for 5 months

karolherbst ,
@karolherbst@chaos.social avatar

@sima at some point you come to the realization that you don't have to do anything at which point the consequences are just a factor in the overall situation 🙃

Lyude , to Random
@Lyude@queer.party avatar

"oh so you want to tell the Democrats that you'll vote for them no matter what?"
listen I get your criticism but I don't follow what makes you think there's going to be an election after 2025 if trump wins.

karolherbst ,
@karolherbst@chaos.social avatar

@Lyude something something that one party needed only one month to kill democracy something something...

karolherbst , to Random
@karolherbst@chaos.social avatar

I have off this entire week and it's already starting great (multiple things happening)

and like... ahhhh....

karolherbst , (edited ) to Random
@karolherbst@chaos.social avatar

Hades status:

Current Streak: 5 (and I always use the +20% weapon)

Apparently I haven't actually completed the first game when I started Hades 2 🙃 But I think I played a Hades initially when the story was different than in the current version? At least some of the things happen in a different way than I remember.

cas , to Random
@cas@treehouse.systems avatar

sleeping pattern go brrrr

kinda fun to get these calendar entries lol

karolherbst ,
@karolherbst@chaos.social avatar

@cas oh that's me this week....

karolherbst ,
@karolherbst@chaos.social avatar

@cas maybe I should start tracking it as well, because I'm kinda curious how much I actually sleep.

What's like a cool/reliable-enough smartphone based solution here?

karolherbst , to Random
@karolherbst@chaos.social avatar

I should check out this 140 thing everybody is been talking about lately 🙃

karolherbst , to Random
@karolherbst@chaos.social avatar

git shouldn't disclose any email addresses or rather, we should just remove email addresses altogether from git

karolherbst OP ,
@karolherbst@chaos.social avatar

@ljrk yeah, though in git it's just weird, because there are still important projects using email based workflows and ahhhhhhh

karolherbst OP ,
@karolherbst@chaos.social avatar

@Man2Dev sure, but an email is entirely irrelevant to the hole signing stuff thing.

Sure it's based on gpg now, but it totally doesn't have to, and it's also entirely irrelevant which email address an identity uses here either.

Could also say "email makes sense because we used to use mailing lists", but the point is: most don't and having an email there is kinda besides the point where people really use other means of communication these days.

karolherbst OP ,
@karolherbst@chaos.social avatar

@neobrain wait what? 🙃

karolherbst OP ,
@karolherbst@chaos.social avatar

@aitorpazos I'm not saying there shouldn't be a way of communicating. But like the email doesn't have to be in the git repository.

If you use a git forge you still have an account and can enforce whatever policy you have, but this way you can e.g. make email only visible to maintainers of the project and things like that.

And if somebody doesn't want to be reached by email that's an entirely fair request and should be respected.

karolherbst OP ,
@karolherbst@chaos.social avatar

@raimue that's like entirely irrelevant for code in a git repo?

karolherbst OP ,
@karolherbst@chaos.social avatar

@raimue yeah, but then we can just remove it from the repo and link to the account identity in the git forge.

And if projects really want to require emails, then they can just tell users to add it to there git user.name thing.

Just if your workflow is all on a git forge and not via email, then the CLI tool saying "but you have to set your email" might make people disclose their email even if they don't really want to and even if the project doesn't care at all.

karolherbst OP ,
@karolherbst@chaos.social avatar

@Man2Dev I don't see how it's relevant where you are using your email elsewhere honestly.

For all I care a single person can use how many identities they like to contribute to projects. What matters is that it's the same identity doing it in the future.

There are other ways of signing content which isn't tied to an email.

The main reason git is doing it via email is, because when it was written, the email workflow was basically the only relevant one at that time.

But now we have 2024.

karolherbst OP ,
@karolherbst@chaos.social avatar

@aitorpazos Why is the email relevant in the first place?

As a maintainer all I care about is to communicate on submitted code.

After the code is merged, it's my responsibility as a maintainer to react on bugs/issues created by code from other/new contributors.

I might reach out to the original contributor, but I don't have to and if they were one time contributors I'd have to fix it myself anyway.

And you usually know how to reach out maintainers, even if it's not in git.

karolherbst OP ,
@karolherbst@chaos.social avatar

@Man2Dev I think you simply overestimate how relevant the email bit there is.

Nobody really cares about linking identity across repositories. And almost nobody cares about signing commits and it doesn't even change a bit besides inconveniencing myself.

Signed tags, I agree, but that's a maintainers job to do, not drive-by contributors.

And the identity part can be part of the signing feature, doesn't have to require an email in every commit.

karolherbst OP ,
@karolherbst@chaos.social avatar

@aitorpazos As a maintainer I own the responsibility whether I squash them or not.

And yes, I see value in people wanting to link their identities, but then again: how is the email relevant here? It could be anything besides email as well. And the git tooling shouldn't make it sound like "it's required" either.

karolherbst OP ,
@karolherbst@chaos.social avatar

@Man2Dev yeah, I'm not disagreeing with signing commits is important in certain areas, I'm just worried that "requiring email", which tbf is more nudging than requiring, is just causing preventable harm, e.g. some person shit-talking some contribution and then the email inbox gets flooded with harassment randomly.

Sure that can still happen on git forges, and probably will, but at least you can still check your emails for actually important things and don't stop looking at it for weeks or so.

karolherbst OP ,
@karolherbst@chaos.social avatar

@Man2Dev right, but we can't expect random drive-by contributors to do this. And people might have started with their private email originally and moved to an alias later.

I'm sure the technical solutions are all there and I'm sufficiently proficient myself and keeping all the mess out, it's just that knowing how to do all that shouldn't be the entry requirement to FOSS honestly.

karolherbst OP ,
@karolherbst@chaos.social avatar

@neobrain why would they even want to reach out to people 🙃 Just file a bug against the project...

karolherbst , to Random
@karolherbst@chaos.social avatar

what's like the most light weight (but still not painful) git forge with proper CI capabilities?

CI as in, I want a job to run regularly and verify a docker-compose setup still works after updating the used images and notify me if it didn't update anything.

Or rather, if git based CI isn't the best solution for that, what would be a better solution?

karolherbst OP ,
@karolherbst@chaos.social avatar

@forestjohnson maybe it would be better to update whenever a new version is there and just verify it works and if not, just roll it back?

Could be easier to do, because then I just need to target the live system with some curl commands to check if everything comes up and if not, I just roll it all back.

There is no state anyway, because the docker-compose I have here is simply deploying reverse nginx proxies. But my Vaultwarden is proxied there so I don't want it to break randomly

astrid , to Random
@astrid@fedi.astrid.tech avatar
karolherbst ,
@karolherbst@chaos.social avatar

@astrid the modern solution to this problem is obviously to have an AI knowing how to invoke help for a certain tool and....

karolherbst , to Random
@karolherbst@chaos.social avatar

how can I port forward traffic directed at a dynamic IP?

Like e.g. eth2 has a specific IP, but it could change over time, how would I be able to declare that in iptables?

Or more high-level: how can I redirect all internal traffic directly targeting my router at a specific port to another port without having to hardcode an IP address

karolherbst OP ,
@karolherbst@chaos.social avatar

@psy I want to catch everything coming from internal or external (wan and lan in openwrt) targeting my external IP and port forward it (e.g. 443 -> 8443).

My openwrt setup is silly and the image force enables port 443 on the web frontend which I want to get around.

Maybe a firewall rule ain't the greatest idea here.

karolherbst OP ,
@karolherbst@chaos.social avatar

@Man2Dev yes, but I still have to set up a firewall rule for reasons not important here.

connections to external IP at port_a -> somewhere else at port_b

karolherbst OP ,
@karolherbst@chaos.social avatar

@Man2Dev yeah, but I'm hosting something at home, so that's why I have this problem.

karolherbst OP ,
@karolherbst@chaos.social avatar

@Man2Dev sure, but this is all running on my router and I need to redirect stuff hitting my router, because it's the target of my external IP, but it also starts a service at port 443, which I want to get around.

karolherbst OP ,
@karolherbst@chaos.social avatar

@Man2Dev sure, but how would that help if my phone targets my external IP at port 443 from inside my home network, if the service I actually want to target does not run on my router at port 443?

karolherbst OP ,
@karolherbst@chaos.social avatar

@Man2Dev yes, and for port forwarding I either have to specify a fixed target IP (which is dynamic, because it's my external IP given by my ISP) or redirect all traffic (breaking internet access)

karolherbst OP ,
@karolherbst@chaos.social avatar

I found a different solution where I mess with the package manager to prevent it from writing a file breaking my setup 🙃

karolherbst , to Random
@karolherbst@chaos.social avatar

today's project: running docker on my router

karolherbst , to Random
@karolherbst@chaos.social avatar

I'm surprised how quickly we went from "companies of communication tools aren't to touch any internal communication for any purpose of enterprise customers" to "but for AI that's perfectly reasonable!"

Lyude , to Random
@Lyude@queer.party avatar

slack: let's train a model on everyone's conversation
ignoring the fact I uh, am not at all clear on what you even be training a model for here, don't a bunch of corporations filled with confidential information they think is REALLY important pay you large sums of money
like it's one thing to steal users data, like yeah that's bad but you know corporations will just keep doing it and not care, but uh. slack I think you're stealing other corporations info here and they have money, a lot of it, a lot of it which you depend on coincidentally

karolherbst ,
@karolherbst@chaos.social avatar

@Lyude apparently "auto-completion" 🤓

Lyude , to Random
@Lyude@queer.party avatar

not upset with anyone
just
reminder for y'all: please, I'm not interested in systemd discussions. please stop bringing them up with me.

karolherbst ,
@karolherbst@chaos.social avatar

@Lyude wait.. people approach you to talk about systemd?

karolherbst , to Random
@karolherbst@chaos.social avatar

once again, why are tech-bros

Lyude , to Random
@Lyude@queer.party avatar

Can anyone else running spotify on Linux confirm if the spotify flatpak is now broken?
Launching it no longer works, it just keeps spitting this out in journalctl:
May 16 17:47:25 emerald org.gnome.Software.desktop[4288]: /app/extra/bin/spotify: /usr/lib/x86_64-linux-gnu/libcurl.so.4: no version information available (required by /app/extra/bin/spotify)

karolherbst ,
@karolherbst@chaos.social avatar

@Lyude works here, but the flatpak was last update a month ago or so?

karolherbst ,
@karolherbst@chaos.social avatar

@Lyude I get the same message though and I would be surprised if that's the reason for it not working for you

sludgegirl , to Random
@sludgegirl@chaos.social avatar

Why the heck are we so sensitive to electrical noise all of a sudden, it feels both absolutely silent and completely deafening

karolherbst ,
@karolherbst@chaos.social avatar

@sludgegirl I hope it has nothing to do with me talking about it recently 🙃

karolherbst , to Random
@karolherbst@chaos.social avatar

why is summer...

karolherbst , to Random
@karolherbst@chaos.social avatar

I wonder if #rusticl is mature enough to be enabled by default on a few drivers these days.

I'm still a bit reluctant, but maybe it's fine?

karolherbst OP ,
@karolherbst@chaos.social avatar

@herrorange depends on the driver honestly. But yeah.. I think I wouldn't have considered it ready before 24.0 myself.

But yeah... can just report those issues and I'd need to fix them, but there are a couple of people having a relatively good experience with it so far.

karolherbst OP ,
@karolherbst@chaos.social avatar

@darktable yeah, so I've heard 🙃 But my main motivation here would be to get more testing, especially while mesa is still in RC, to fix things before they hit the release.

But yeah.. I've just fixed another bug hit with Darktable this week

gfxstrand , to Random
@gfxstrand@mastodon.gamedev.place avatar

This week's project: Reworking NVK cbuf support. We've had a lot of issues with too much internal stalling and I think a lot of them come down to the fact that we're re-binding cbufs every draw call.

My plan for root constants, is to do inline updates with the LOAD_CONSTANT_BUFFER command. I don't know how much of a difference there is but I strongly suspect this pipelines much better.

For bound cbufs, I'm planning to just make our dirty tracking way more competent.

We'll see how it goes!

karolherbst ,
@karolherbst@chaos.social avatar

@gfxstrand might be a fun idea to figure out if it's deterministic which thread writes the value.

karolherbst ,
@karolherbst@chaos.social avatar

@gfxstrand sure, but what if only ever the first thread in the subgroup, not the first thread of active ones writes the value?

Then it still wouldn't help, but at least we would know what the rules are.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • random
  • haupteingang
  • All magazines