Notices by Brian Swetland (swetland@chaos.social)

@whitequark @mcc That definitely aligns with my experience trying to read it from a "how do I build a functional WASM runtime" and "how do I generate correct WASM binaries" standpoint.

Some notes from my initial encounters with it illustrate specific grumbles with the presentation:
https://chaos.social/@swetland/111271568378452789

@mcc @whitequark Between that and how just incredibly convoluted the spec is (any needed piece of information on an opcode or how it works is inevitably spread between 3-5 different sections of the spec), WASM really makes my head hurt. There are some good ideas in there and people are doing good things with it, but from a "sit down and understand how it all works" standpoint it's a pain in the butt.

@whitequark I've been meaning to check this out now that the various single-pair ethernet PHYs are in production. Especially curious how complex/expensive the PoE options are because combining single-pair PHYs with little MCUs sounds like a lot of fun especially if you can deliver power too...

@whitequark Optimizing these debug transports is fun. Looking at this comment in my SWD tooling from ages ago (about how many SWD ops fit in command buffer to the probe firmware)...

// 10 txns overhead per 128 read txns - 126KB/s on 72MHz STM32F
// 8 txns overhead per 128 write txns - 99KB/s on 72MHz STM32F

@whitequark it is a mystery!

@whitequark I think (if I'm remembering the layout relative to the JTAG scan direction correctly) that this "design decision" falls out of wanting to be able to get at the DBGBREAK bit by itself at times (avoiding scanning in/out 32bits of extra overhead).

@whitequark SWD has its warts, but it feels mostly (sometimes over-) designed... whereas the ARMTDMI stuff feels very, uh, tacked on after the fact is probably kindest... then ARM9 (not A9)... well I've forgotten the specifics, but I what remember is it being an even more convoluted evolution.

@whitequark Having done ARM7TDMI JTAG debug interface implementation first, back in the day, made the convoluted bits of SWD based debug really seem not all that bad by comparison.

It's been long enough now that your posts on this are bringing back more nostalgia than trauma.

Am I imagining things or didn't there used to be a way to attach spoiler-protected images to posts without putting the entire post behind a CW?

#Mastodon #Questions

I hate a lot of things about "AI", but one of the most infuriating things to me is people responding to questions (in chat systems, social media, forums, etc) by cutting and pasting ChatGPT or whathaveyou.

Please, if you don't know and can't be bothered to actually find out, just don't reply at all. Thanks!

I mean, seriously, if I wanted the shitty "AI" answer I could have just asked one of any number of shitty "AI" agents out there. You are not helping here.

@mcc Strong agreement with this assessment.

Similarly, they have open protocols and open source implementations for the backend, but as @cwebber 's excellent writeup points out, operating a full independent instance is nontrivial, and lacking any actual full instances users could migrate to, the promises of being able to "take your account elsewhere" remain mostly aspirational for the time being.

I mostly hope that fedi/activitypub is inspired by some of the interesting bsky features...

@AzulCrescent The worst ones are where even the bowl is an add-on and the default experience just gives you a handful of flour.

Hey, #ActivityPub folks.

I'm curious -- has there been thought given to making some kind of simple HeyFollowMe button on web pages that could work as smoothly as RSS buttons did for RSS using folks back in the day?

Ideally not requiring one to copy and paste a URL into a #Mastodon search box or type your home server address into a Follow dialog box on another site's page.

I'm assuming there's some reason such a thing isn't already out there.

#Question

@mikolaj @msw Maybe they're thinking about whatever happens with the inevitable fork(s) of the pre-"source-available"-licensed version of Redis...

@thomasfuchs @anildash I'd potentially be interested in AR if it could be delivered in a form factor not much different from the actual glasses I already wear (as opposed to something on the ski-mask to motorcycle-helmet spectrum) with sufficient resolution, battery life, integration, etc.

Of course the killer use-case of "remind me who these people are I'm talking to" immediately falls into a huge pit of privacy and invasiveness concerns for everyone on the other side of the glasses.

A thought: Which is the last generation that remembers computers that just hard crashed all the time?

Where you learned to Save Often and Make Backups because even if the App worked great, the whole machine might just blow up at any moment.

I feel like, for me, this pretty much stopped being a thing by the early 2000s. But I also started using Linux as a primary OS in the mid-90s, so mostly saw Win95/ME/etc from afar.

I do remember when X11 would sometimes just explode, killing all my apps.

@mos_8502 I'm doing this for my little systems language compiler project as a bootstrapping mechanism -- a minimal transpiler that generates C from a subset of the target language so I can then write the full compiler in itself.

Previously I was writing a full compiler in C with the thought to do a mechanical translation step after, but as I started digging deeper into optimization, etc, I was unhappy with the complexity of both the code and the translation process. Thus this new approach.