Show newer
Walter boosted

an alternative MNT Reform motherboard with SMARC carrier, would that be interesting?

Walter boosted

@alcinnz, cc: @devinprater, for an alternative, check CMU's . They've got several OSS projects, all with better sounding voices:

* Flite or festival-lite, a small fast portable speech synthesis system.
* Bard Storyteller: an ebook reader that uses Flite.

Both based on the CMU Wilderness Multilingual Speech Dataset, a speech dataset of aligned sentences and audio for ~700 different languages, but it's not licensed and served from a non-HTTPS site:


Walter boosted

Here's the thing, the fancy-looking unicode characters almost never create words on their own, especially not those from the web, gemini or IMs, nor the ones for STEM fields.

A rudimentary look ahead "algorithm" can toggle transliteration on and off. We have transliteration mappings in every web CMS that deals with user uploaded files, because they need clean paths for SEO, just take the mappings from Drupal.

cc: @lena @aral @devinprater

Walter boosted

re: defining "simple" software (subtoot) 

This is not an arbitrary list of tasks. For more information, see NASA NPR-7150.2D "Class A" software certification requirements.

There are many safety-critical software standards out there, depending on country and industry, and they all boil down to the same thing: designing a nice, sanely architected and abstracted software stack, is not the same as designing a simple software stack.

Abstractions tend to increase software complexity by making complexity easier to write.

Show thread

Lewd programers arguing about FatELVES. That's ELF, but plural. 

Ryan Gordon Halts FatELF Project by Thom Holwerda on May 5, 2009.

P.S. Sorry, somebody had to put it here on the fediverse.

Walter boosted

the culture we share here is going to change the world forever.

long after the fediverse is offline, and 99% of all our data here has rotted away, people will look back and wonder how we could have shared something so beautiful.

maybe this sounds way too overblown, but as i sift through bitsavers over and over again, i am astounded every time at the (now forgotten) efforts people made at trying to extend computer science in ways which was friendly and pedagogical.

what we have now in computing is a pale, dim shadow of the likes of DEC, Smalltalk, Lisp machines…

6) the developer tools are practically nonexistent.

All these extra issues might sound irrelevent in the short term, because apps need to be accessible, but without a solid foundation you can't build anything.

So, new POSIX proposal for extra file descriptors, and a way to process the structured data and send it to other hardware or software for improved accessibility and also to the current stdout, if needed.

What have I gotten myself into?

Show thread

To recap:

1) no screenreader support at startup

2) nothing inside the firmware settings (≈BIOS)

3) in GRUB, from what I've read, you get some beeps in one distro and you could maybe add something extra on top, but forget about encryption passwords, there's nothing hide here.

4) in most cases you get the screen reader after login, resolvable.

5) everything else is painful to use, even CLI tools, because they work exactly like those controlled by telegraph typewritters.


Show thread

The state of linux screen reader support seems to b3 totally absent at the lowest levels and embarrasing in userspace.

I need some simple low level interfaces for creating a framebuffer-like device for audio, plus a decent TTS engine.

I was hopping for something in coreboot. But there's nothing useful there and, to rub more salt into the wound, coreboot exposes a complex audio interface for platform independent drivers with a full specification costs a fortune in OSS money


I need an extra standard stream in POSIX to use with stdout, but with binary data (Cap'N Proto, but simpler) in order to avoid the need for date and number formatting or table layouts.

This extra stream would contain all needed for the presentation layer, it will help both users that require assitive technologies, but also when opening child commands, because you no longer need to parse stdout. No, JSON or BSON are too much.

Also RFCs ask that you error out and don't try to guess encodings.

Show thread

Does anyone know why we have exactly 3 standard IO streams (stdin, stdout, stderr)?

It's not like we still need to use real TTYs.

We could have used at least a few extra inherited streams, especially when using pipes or opening child processes. I need one for improving and security.

There once was a proposal to ad an extra opt-in file descriptor for defending against attacks, that's all I found. But the reference is broken, and it's late here.

Walter boosted

A really good forum post showing why developers should just listen to blind people, especially blind developers, and not try to re-invent any wheels. Mudlet accessibility #a11y

Walter boosted


if something is done in a way that seems weird to you, it's a good idea to research why it's done that way because sometimes things are done in certain ways because they have to be and you're not the first fucking person in computer science to discover the complex concept of memory allocation or static linking

Show thread
Walter boosted

@alcinnz @pettter @z428 @humanetech Oh, and one more thing, these standards need to be approachable by individuals.

Which means these standards need to be written to be very clear, and must be small enough to be understood by individual contributors.

For example, I just searched for "open document format specification". The results are a pile of irrelevant sites that are happy to explain what ODF files are at an abstract level, but none provide the specifications. There is a single link to towards the bottom of my search results.

I click it, and I get a wonderful brochure site that is happy to sell me the idea of adopting ODF for my document processing needs, but again, no specs.

I click on Developers, and I am once again presented with a literal pile of irrelevant bullshit. There is one statement in the first paragraph which tells me, "Sorry, Mario, but the princess is in another castle!" In this case, I have to go to some place named "OASIS".

I click on OASIS and I'm lead to a technical committee description page. Still no specs.

At this point, I'm giving up.

But, apparently, ODF is a proper ISO specification now, so I now try that route.

After some rummaging around on the web, I finally land on this:

By george, it's an actual spec. But holy fuck, it costs me almost $200.

Yeah, it is "open in name only, for it is neither."

Contrast this with RapidIO specifications, a major competitor to HyperTransport and QPI bus interconnects on high-performance microprocessor-based systems:

Go to
Hover over "Technology" and click on "RapidIO Specifications"
Click on any PDF link your heart desires. THOUSANDS of pages of specs, at your immediate beck and call. Have fun.

Walter boosted

generative art, hacking the gibson 


Prompt: inspirational poster about hacking the gibson

The AI decided Snowden was a great fit for this one? I really like this one.

(I’ll reply with the other generations)

I'm too tired for this B.S., I really hope that I've missed something and that INTEL-SA-00617 doesn't really say that my 3rd generation i5 is abandonware.

Show thread

PSA: Your CPU microcode doesn't just Meltdown and Spectre, it's now abandonware! (INTEL-SA-00617)

July 17, 2022 Qubes Security Bulletin 081:

"An adversary who controls a VM with an assigned PCI device can infer the memory content of […] Our best evidence indicates that mitigations are available for all and only those CPUs that are both eligible for and updated with the Intel microcode update released in May 2022."

Show thread

Abandonware in phones, fridges, toasters, TVs and now CPUs?!? You got to be kidding me! 🤦‍♂️

Walter boosted

Note: Terminals are shit, and the standards are made up and you need to wade through literal decades of cruft that no user will ever hit.

I don't blame anyone for being bad at this. I'm terrible. I just assumed other people were better.

Show thread
Walter boosted

One of the most significant papers from Japan's sample return mission to asteroid Ryugu was published in Science today.

It's a monumental result, according to other planetary scientists. Essentially, we now have access to pristine materials from the very earliest days of the solar system, some 4.5ish billion years ago and we can study them on Earth.

A big deal.

Show older

The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!