Posts
4578
Following
317
Followers
476
Linux kernel hacker and maintainer etc.

OpenPGP: 3AB05486C7752FE1
Storj is a resource (or memory) manager. All software I've worked in the past from drivers up to applications ends up converging into a resource manger. Once you can nail the resource manager use case, the world will be eventually yours. It's simple as that.

Wolt, Ueber and Spotify are also resource managers and not much else. For "Spotify" in particular the latency is not there yet but Storj gives the metric that reaching that scale is only few years away.
0
0
0
If you market random tokens to end user it is exactly like preaching about IP adresses.
1
0
0

Jarkko Sakkinen

Storj made me consider a blockchain company. Before that, honestly I did not understand the application. I used it for four months without knowing it is token based and ended my Dropbox subscription. Token is like IP address, totally useless without service that makes it tick. In Storj tokens work as B2B in storage sharing. I just pay with my credit card and enjoy the awesome benefits like pay for use and ultra-cheap S3 object storage, never ever thinking crypto 😀 This is how to do it right.
1
0
0

Jarkko Sakkinen

Edited 1 year ago
... which is not same as saying systemd sucks overall OR I would some how "generally" dislike it. It is great in those deployments where it fits (e.g. my home file server and desktop PC) ;-) </disclaimer>
0
0
1

Jarkko Sakkinen

Edited 1 year ago
#systemd sucks for #embedded because it is not a design that downscales effectively. Tried systemd few times for smaller devices but it just a bad solution for anything not at least a mobile phone.

An init system that could both upscale and downscale and based on statically linked tight core would be awesome.

It could even be (partially) compatible with the same unit-file format.

In particular for this project it would make a lot of sense to implement such support because this is a need at scale:

https://github.com/uutils/coreutils

Could even gain sponsorships from embedded companies.
1
0
4

Jarkko Sakkinen

I have now a registered private company for the sake of being a contractor starting from October: Siltakatu Solutions Oy (aka Ltd).

Siltakatu is my home street [1] in #Tampere and I try to find solutions for various issues in software engineering, so by combining these two facts I got a company name 🤷

Sometimes my solutions suck tho but I do not imply anything about quality in my "company brand" ;-)

[1] The #Finnish word silta translates to bridge and katu translates to street.
1
0
1

Jarkko Sakkinen

Edited 1 year ago
One wisdom that applies both understanding math and programming in deeper level and navigating through that jungle: be stupid and try not to understand 🤷 Stupid is the new smart.
0
0
0

Jarkko Sakkinen

Edited 1 year ago
A power user belongs to a category of users who use computers just for the sake of using them, and that is the overall main goal of using a computer.
0
0
0

Jarkko Sakkinen

Edited 1 year ago
I like Low Level Learning Youtube-channel but with this I have I have to totally disagree: https://www.youtube.com/watch?v=3T3ZDquDDVg

The golden rule for writing code and debugging is obviously:

1. Do the most simplest and least scalable trial at first.
2. If that does not scale take a more powerful tool.

I certainly know how to use core files but it is MORE EFFORT.
0
0
1

Jarkko Sakkinen

Edited 1 year ago
The way I see it how blockchains are is like early 90s, 9600bps modem and dial up. One practical issue that they might perhaps solve is mapping the dark matter of revenue streams that are ”below two decimals”.

I think their key application is a "pay for utilization" cost model, instead ”pay for subscription time” as an option. I've used a S3 storage service Storj for some time and for my use this has been quite effective.
0
0
0

Jarkko Sakkinen

Starting from October, when my research contract ends, I'll be joining Parity as compiler and virtualization engineer: https://www.parity.io/
1
0
1

Jarkko Sakkinen

Edited 1 year ago

Since https://github.com/RustScan/RustScan claims to be fast I tried first a trivial SYN scan to my file server:

sudo nmap -sS -p- haaparousku

I got the list of ports in about one second (bit more or less). Then I tried rustscan and got bored enough to finally SIGINT it:

rustscan --addresses haaparousku

Not sure I get the improvement here… Also command-line is not “tactile :-)

Something definitely to improve in Rust command-line apps is not to think arguments as an object tree that you feed into clap crate. That’s just lazy.

Instead a better metaphor for a great command-line interface is something like a game pad that you can “play” easily. That’s why nmap has been relevant for decades.

0
0
1
Edited 1 year ago

Glue on pizza is apparently not enough, how about bogus AI mushroom foraging books
https://www.reddit.com/r/LegalAdviceUK/comments/1etko9h/family_poisoned_after_using_aigenerated_mushroom/

1
1
1
I guess I could create my own XML parser which then outputs run-qemu.sh.
0
0
0
I even used to bootstrap qemu and swtpm for every single build at one point ;-) It is great when testing kernel features when you can always start from known state.
1
0
0
@stsquad I generate the whole operating system for each kernel test build. I don't have a machine description when I start compiling it.

Essentially each test VM is also one shot, deleted after a single test run.

Basis is like https://codeberg.org/jarkko/linux-tpmdd-test
1
0
0
@janotomko It would be easier to have a few template configurations and XML can be more easily post-manipulated.
0
0
0
Show older