@janantos Has been a while when I visited in a library :-) It's a good tip anyhow, I'll go check. Thanks!

@aks Not pointing out fingers here, and do not even try to change this madness but at least I have right to complain and whine ;-) Making sure that an open source project somehow linearizes its history of communication is an incredible asset and definitely worth of pursuing for. I would do my job factors worse if e.g. kernel ML archives and lore.kernel.org did not exist. At min once a week I'm looking for some old discussion or similar just to check up on things.

You don't have to do this or make effort on establishing such ecosystem, that is true, but I'd look it also as an competitive advantage to have a documented history of work done.

@aks The features you described are what I would consider harmful features [1]. In the past there was thing called a meaningful chat history, which could be interpreted and studied years and years after the discussion took place.

IRC is also the only protocol of which raw protocol dump is easily interpretable without help of a machine. XML/JSON protocol dump is waste or human interpreting it manually can be claimed to be a subject to brutal torture :-)

I'm not capable on understanding what is going on in a typical workplace chat overall even for the span of a single day because they are full of videos, code snippets and countless other distraction signals. By any practical means, they are usable only in that moment when the messages are being exchanged. It's literally like having a workplace discussion by using the walls of a public toilet as the main medium for discussion.

IRCv3 adds some of the modern ideas but not at the cost of retainable chat history.

[1] Thanks to Github who created this culture of bad quality. Steady stream of bad and worse chat services such as Slack, Discord and many others followed the lead.

We don't have enough kernel memes on here, just doing my part

What is this:

https://opensource.org/ai

"Open Source" is "Open Source AI"

OK, cool.

This must be then an AI:

```
int main(void) {}
```

For me this has an appearance of a scam, and it also reflects a recent outbursts such as e.g. this one from Microsoft:

https://www.techspot.com/news/103609-microsoft-ai-ceo-content-open-web-freeware-ai.html

Not pointing out to Microsoft here in particular. It was just so arrogant, obnoxious and ignorant take from Mustafa Suleyman that it carved to my brain forever.

IT giants have put way way way too huge stakes on top of the AI game board. There's also a strong motivation to compound "Open Source" and "AI" as interchangeable concepts to justify evil corporate tactics. Right from the get go OpenAI picked up a name, which misleads by implying "openness" (with no anchor to anything actually open).

Standards share no resemblance to this eye roll trick. I read them a lot as part of my work (the most recent was DWARF4 spec few weeks ago), and I know what a standard should look like, if anything.

This is not an opinion about AI way or another but I recognize a fake standard 100% when I see one, and this is as fake as it can ever get. It could be at worst interpreted as an attack against open source and free software governance, and the values that we believe in.

RT @fsfe

Cheap #3D #printing services in the #EU region? I just need to print one a spare part that is not complex and #ThingVerse is feasible for acquiring the model.

E.g. I get room keys as a security feature. I did read the Matrix spec. But IRL I find myself carrying USB stick from computer to another in order to have anything readable if a room is encrypted. I.e. I can either use recovery key or authorization request to activate a client but it gets you only hafl-way there.

This makes me feel that UX was designed to fit the security features and not other way around. And if you talk to a Matrix cult member they will say to you that you're not getting it.

I miss the 90s, IRC and Polish hacking groups bombing IRCNet with regular netsplits :-) And who needs security when you have OTR (innovation from 2004).

In software security a protocol built without security is easier to guard than a protocol with "security features". IRC has this right too. I feel that this some annual thing to praise the IRC and hate the world 🤷

I need to visit local museum and go to see the first IRC server and ask for guidance in this misguided world (the first ever is 400 m away from my home door, i.e. living at a holy place).

After having to use Matrix for a while and seeing how bad Slack and Discord are, I wish there was a criminal law that would allow IRC only for IM. Steady performer, zero fucks. #IRC

The CFP for the Containers devroom at @fosdem 2025 is now open! https://discuss.linuxcontainers.org/t/fosdem-2025-containers-devroom-call-for-papers/21956

After four weeks doing Rust it became clear to me that LSP is quite useful sometimes, when you have to deal with a language that was not designed to be written by humans.

[That said: even less so by Copilot, Bob or Clippy]

;-)

I've managed mostly with vim and rusty-tags so far but to help with code exploration I installed vscodium. Seems to do the job since it has a vim emulation plugin and works together with rust-analyzer. And LSP does not work too well in terminal anyhow...

PS. This is the most useful documentation that I found on topic: https://bootlin.com/pub/conferences/2020/elce/opdenacker-using-vs-code-for-embedded-development/opdenacker-using-vs-code-for-embedded-development.pdf

The end result:

$ df -h
Filesystem      Size  Used Avail Use% Mounted on
/dev/sda1       932G   14G  918G   2% /
vendorfw         16G   79M   16G   1% /usr/lib/firmware/vendor
devtmpfs        4.0M     0  4.0M   0% /dev
tmpfs            16G     0   16G   0% /dev/shm
tmpfs           6.3G   22M  6.2G   1% /run
tmpfs            16G   18M   16G   1% /tmp
/dev/nvme0n1p5  974M  234M  673M  26% /boot
/dev/sda1       932G   14G  918G   2% /home
/dev/nvme0n1p4  499M  129M  371M  26% /boot/efi
tmpfs           3.2G  864K  3.2G   1% /run/user/1000

$ sudo btrfs filesystem show  
Label: 'fedora'  uuid: ff4c6159-93b4-49db-b9fe-13fdb502563c
	Total devices 1 FS bytes used 12.33GiB
	devid    2 size 931.51GiB used 20.06GiB path /dev/sda1

I installed Asahi Linux to a 128 GB rootfs (out of 1 TB of internal storage). It is connected to a dock with an external 1 TB M.2 drive. Fedora sees this drive as /dev/sda.

Migrating the subvolume of /home was just a matter of creating a single all encompassing BTRFS partition to /dev/sda and then:

sudo btrfs device add /dev/sda1 /home -f
sudo btrfs device remove /dev/nvme0n1p6 /home -f

Now there is an airtight separation / and /home physically, and also they pool space only within their own cages.

Apple firmware supports only booting macOS from external storage, thus the rootfs must always reside in the internal storage, but I think this is already quite sustainable way to deal with it.

This bottleneck/policy can be mitigated but it does not pay the price. It can be overcome by installing macOS to an external storage and consuming internal storage in its full extent for Asahi Linux. Just mentioning this last for completeness ;-)

Encryption can/could be done by following https://davidalger.com/posts/fedora-asahi-remix-on-apple-silicon-with-luks-encryption/ with “a twist’ (since different partition/drive topology).

#apple #macos #linux #asahilinux

US Ambassador to Finland's immediate reaction of Trump's win in 2016 while still serving 😅 #Trump #diplomacy

my friends have started to bully me as "jarkko.js" (those are my name letters). did not came to mind when picking username for my github account duh...

@kernellogger ;-) https://gist.github.com/jarkkojs/d95bbfaf9b35a18fc9199ab72e51a91c

It's a start.

Since I guess Google knows Rust better than me I guess I'm doing the state of art with custom target JSON-files, i.e. deploying them to a random directory in the project directory :-)

I kind of hoped that there would be a place (let's say .cargo/targets or similar) under the project that gets automatically recognized by the tooling. Then you could just put that to rust-toolchain.xml like any other target.

https://security.googleblog.com/2024/09/deploying-rust-in-existing-firmware.html

#cargo #rust #rustup

yamaha piano https://lore.kernel.org/lkml/752542d4a8ff0b73a16a92002bae6338eb87d88c385584918bbc005e4e617b10@mx.google.com/T/#u

@kernellogger I'm also going to try out your instructions, try to polish them and come up with something :-) I'll share the results when I have something. Compiling kernel is my thing (I guess) ;-)