Apart from the obvious duties of a graphical display device, TerminalThing would also be useful from the point of view of dystems security, authenticating users on behalf of the main CPU ( or others). Only after the user is authenticated will TerminalThing be allowed to send data to or request data from the main computer component. For example, there will be no way that TThing can send mouse or keyboard input to or before a user is authenticated.

@h boosted

Tor users: We need your help!

What do you use Tor for? Why do you need it? What has Tor done for you? What could have happened if you weren’t able to use Tor?

Here's how you can share your story with us: blog.torproject.org/how-has-to

@h boosted

Man I could work on getting the #ActivityPub #golang implementation to v1 but there's #Oktoberfest this weekend in Munich and I hear the beers there are as big as my face. So mixing the two is definitely not an option.

@h boosted

The undeniable retro-modern appeal of cyberpunk aesthetics are a vector for messaging and culture that must be possessed and maintained by we freaks, lest they be used against us.

@h boosted
@h boosted

the answer that occurs to me is to hash the parameters and use the resulting number to seed a random number generation algorithm. but then I need to make sure that either (a) similar seeds for the RNG don't create similar sequences of random numbers, or (b) the hashing algorithm produces really well-distributed values, even for similar inputs.

oh and this should be fast (like something that you can do tens of thousands of times a second with cycles to spare)

Capitalist sssholes keep driving people away replicating their tried and true models of conquest and domination in the name of prosperity a (for some).and "diversity". Centralisation is what caused the problem and only bad actors and fucking idiots will tell you otherwise to keep benefiting from the work of others.

You can't do better because you are the problem, asshole.

Modern solutions for some of these problems (like the elephant photo) have become more apparent in recent times with the advent of publicly available trained models, and ML & DL frameworks becoming readily available. Raskin couldn't have improved on automatic classificarion and tagging around the turn on the century in practice, but it's more clear now (after an explosion of interest in DL in the last decade) how OS agents could achieve this.

Take for example, I quite like the idea that files should 't necessarily have to have filenames. Classifying images is a task that can be done more or less reliably using computer vision techniques. If there's an image file which contains a photo of an elephant taken in Southeast Africa, there is no reason it should have a filename DSQ20170924.JPG

I'm re-reading Jef Raskin's The Humane Interface again.
Some things still feel brilliant as ever. Others felt right many years ago before the mobile uis appeared, and we have seen that a few of those aren't as great ideas as they initially seemed to be.
Other ideas may not make a lot of sense if implemented in the way Raskin suggested they could work better, but aided by more recent developments they could be steered in a better direction.
I should take notes and classify these ideas better.

@h boosted

@vertigo @haitch

Even with funding I think the "when it's done" approach is preferred.

I'm sure we've all experienced what happens when a project is accelerated by throwing money at it...

I'm in a similar boat in that I'm using ARM in #rainpsc until alternatives are ready. That way I'm not blocked and can let alternatives evolve at their own pace.

For me the key is modularity, with a focus on quality interfaces that could work with future configurations.

@h boosted

So very, very glad I got that blog article done when I did.

Just in the nick of time, too. Looks like I have some thinking to do on ByteLink (which will now fragment into two technologies: one for RS-232-ish serial interfaces, and one for 4-bit connections).

Which is just as well; I never liked the name "ByteLink" anyway; it should rightfully be called "NybbleLink", and that's just what's going to happen going forward.

@h boosted

Been spending some time researching "geometric diodes" using graphite and/or graphene. Who ever would have thought a circuit element that **literally** looks like the schematic symbol of a diode .... would have diode characteristics? Talk about prescience!

Another motivation to finish the too -- I'd like to make a peripheral for it which would let me use it as an I-V curve tracer, so I can try to replicate some of these results myself.

@h boosted

The TS100 is a soldering iron with an STM32 microcontroller and published firmware making it hackable. This guy added an oscilloscope feature:


I have an STM32 and installed a #FORTH on it; you see where this is going...

@h boosted

@haitch @enkiv2 @alcinnz @vertigo

I’ve been thinking a lot about your modular approach to the hardware and now I see why you are so focused on the physical interconnect.

The whole thing makes a lot of sense and frees me up to think deeper about my primary components of the system.

@alcinnz I took seriously your suggestion about utf-16 and utf-32 support but this appears to be a discouraged practice in languages and systems other than Python 3.x. In the case of Julia, it once had support for all three encodings, but that was eventually dropped in favour of using utf-8 exclusiely.

There's even an "UTF-8 Everywhere" manifesto. I'm wondering what I'm missing and what would be a scenario where utf-32 and utf-16 are absolutely necessary.

*mental note*
- Write a blog post comparing homogeneous OS/UI systems to ideal self-similar counterparts.
- Describe oneideally self-similar hypermedia system.

Yeah, like anybody needed a Microsoft employee's view on manipulation.


@h boosted

@Felthry @troubleMoney But the internet isn't a big truck, it's a series of tubes.

So, a residential water supply line is typically 3/4" or 1" in the US, apparently, with 40-45 PSI being typical. Let's go with 1" - broadband! - and 45 PSI, and I think that comes out to about 40 gallons per minute.

At 15 x 11 x 1 mm (0.165 cc) per card, 40 gallons of SD cards is about 917,676 SD cards.

At 256 GiB per card, you're looking at 29.87 pebibits per second.

Show more

Follow friends and discover new ones. Publish anything you want: links, pictures, text, video. This server is run by the main developers of the Mastodon project. Everyone is welcome as long as you follow our code of conduct!