Notices by Ignas Kiela (ignaloidas@not.acu.lt)

@sun@shitposter.world Oh, thats essentially hydrogen peroxide, interesting. Stuff I get here is a weird mix of alcohols and oils in which the wax mostly dissolves.

@sun@shitposter.world there is some otc stuff here for clearing ear clogs, check if you can get something like that? Water isn't the best for clearing earwax, because earwax kinda puffs up in water, it can make it harder to clear out.

@toast@donotsta.re I wonder if GNS (the gnu thing) could be a decent way to transition and maybe solve some of the technical issues

I mean, right now not, but maybe if ICANN takes it on, a gradual transition onto GNS could viably happen (even if keeping the structural problems of the current domain name system)

@Hyolobrika@social.fbxl.net @toast@donotsta.re It's still a hierarchy of data. It's just that when you want to get some data, you don't go through a hierarchy of data providers.

@Hyolobrika@social.fbxl.net @toast@donotsta.re it does

I can make a namespace, say it's canonical, and give out portions of said namespace for money.

It does literally nothing on the organizational front, the data is laid out in quite literally the same way as DNS, it's just that for retrieving and managing that data you don't need to rely on central servers. You still need a root for GNS, some canonical namespace.

@Hyolobrika@social.fbxl.net @toast@donotsta.re not any more hackily at all, every zone can be treated as a root zone in DNS as well

It shouldn't be too difficult to replace the traditional DNS root servers managed by ICANN to something different, and you'd get exactly what you do with GNS

@wolf480pl@mstdn.io I have zero clue, but because golang seems to almost always pick the stupidest option available, it's gotta be the error message one

@wolf480pl@mstdn.io that fucking language

I kinda want to transition into more devops side of work, but all of the popular tools in that space nowadays are golang and I can't work with golang for more than a week straight without getting too angry to work

Fucking google and their cargo-culters

@wolf480pl@mstdn.io I do data engineering with a side of making and improving the systems that help us deploy and run our code.

My problem is that when I do work with common devops tools, I tend to go fairly deep with them, and often end up hitting problems that either require me to write golang, or are deeply rooted in problems with golang. I like abstracting shit, and golang fights against it even if it gets wrapped around in a bunch of other languages. I filled https://github.com/hashicorp/terraform-provider-kubernetes/pull/1723 as I was working with CDKTF to parametrize a system, even though everything I wrote for that system was written in Pyhton. I'm now thinking of using Pulumi just because it doesn't basically require me to use golang to write providers. I will sooner rewrite Kubernetes then I ever will make a YAML template for it.

A lot of the shit that annoys me in DevOps tools is there because they were built with golang, and as such the design was done with limitations of golang in mind. But when I go to approach solutions, I don't have the limitations of golang in mind, and they always end up slowing me down.

I understand that maybe some things I want to do aren't really required for most dayjob things, but I think they are hurting the ease of use for software. Like, there is a bunch of software that is essentially made to be deployed on Kubernetes, but doesn't have anything better than a Helm package for it, which is fucking terrible. But the design limits golang imposes hardly allow for anything better than templating for software reusability. Reusable components are almost a pipe dream. You don't need them when you're only thinking about your what your company deploys, because you're not running a ton of different deployments of a single software package. But every company writing their own deployment for the same software with mostly the same operating constraints in all cases is fucking wasteful, and just sucks on the industry as a whole. There's businesses that just deploy some open source software in the cloud for you without you needing to worry without that much additional functionality and that's worth paying double the price of raw compute it's getting deployed on because the deployment and operational experience of said software is abysmal, purely because everyone needs to cook up their own deployment from scratch.

I know that it's possible to do basically anything without touching golang, but I find it very annoying when I have stoop down to the level of golang to do things when they so easily could be done in a much better way.

@shibao@misskey.bubbletea.dev 22 bits is tiny

~400 bits is doable with regular computers these days

@azonenberg@ioc.exchange @gsuberland@chaos.social @whitequark@mastodon.social no, T1 are relatively recent as far as ethernet standards go, only came out in 2015 or so

@niconiconi@mk.absturztau.be @feld@friedcheese.us @lanodan@queer.hacktivis.me
Famously there's also 5$ 100G Intel optics available on ebay that most people think come from facebook https://www.ebay.ca/itm/234826973241

@wolf480pl@mstdn.io LPDDR5 does give a fairly good perf boost from better clocks, so makes sense in that regard

Will see if LPCAMM can actually succeed, it does restrict the layout quite a bit, but maybe it will get adopted enough.

@wolf480pl@mstdn.io 16GB is enough tbh

my work laptop has 16GB and it's never been a big issue

now, my work isn't compute intensive on the laptop, but that can be said for at least 80% of office workers

@feld@friedcheese.us @lanodan@queer.hacktivis.me @niconiconi@mk.absturztau.be I understand where it would be useful. I don't think these cases are all that common though (it can often be cheaper to put up a bunch of 10 port switches all over the place than to wire a ton of copper from a central location).

@feld@friedcheese.us @lanodan@queer.hacktivis.me @niconiconi@mk.absturztau.be ix industrial isn't all that cheap either, SFP modules do cost, yes, but ix connectors are currently priced for "industrial" applications, so with a premium.

@feld@friedcheese.us @lanodan@queer.hacktivis.me @niconiconi@mk.absturztau.be honestly kinda doubt it, I think that port density on switch-gear for 10G or less nowadays is not that important.

@feld@friedcheese.us @lanodan@queer.hacktivis.me @niconiconi@mk.absturztau.be right, that's the one I mean when I talked about the ones coming from industry. This one is included on MNT Reform Next. But the laptop will also come with a dongle from it to regular ethernet, because there's zero chance that people transition to using it rather than modular 8P8C

@lanodan@queer.hacktivis.me @niconiconi@mk.absturztau.be I mean, it's a connector design from the 1960's, we learned a lot since then about connector design.

It's just that replacing it is damn near impossible. There are some stuff that's coming from more industrial automation side of things that is getting semi-standard there (because there the brittleness and everything else really sucks), and the new MNT laptop uses one of those by default because it's smaller and sturdier, but I don't really see it coming to the mainstream.

@lanodan@queer.hacktivis.me @niconiconi@mk.absturztau.be fiber isn't expensive tbh, just annoying

it only gets expensive if you want to fiber splice stuff, but even then, basic machines are ~400$ on aliexpress, not unreasonably expensive

you can tell that I've been tempted