fluffysoft

soft as in gentle and pliable
hobbyist and professional maker of many kinds of things. Non-exhaustive list of such things: video games, software tools, radios, antennas, fabrics, yarn, keyboards, wine and spirits.
Been around Merveilles since 2021

pronouns: he/they
site: https://fluffysoft.xyz

Public Key

npub1wsv6mtn6s46yutzt80kqllvl5fsj4t5dgyns79dp03uu8cezw7ssv9klxg

NIP-05 Address

fluffysoft_at_merveilles.town@momostr.pink

Profile Code

nprofile1qqs8gxdd4eag2azw939nhmq0lk06ycf246x5zfc0zkshc7wruv380ggpp4mhxue69uhhjctzw5hx6egpp4mhxue69uhkummn9ekx7mqjqs55c

Publishing to

yabu.me

nos.lol

relay.damus.io

Author Public Key

npub1wsv6mtn6s46yutzt80kqllvl5fsj4t5dgyns79dp03uu8cezw7ssv9klxg

Show more details

Last Notes

2026-10-11 18:12:36 UTC

- reply

I keep coming back to these ideas and adiabatic circuits every couple months. My favourite fantasy in this realm is little bots that wait for a problem to solve, then take that problem with them as they crawl into a bath of liquid nitrogen so they can superconduct while they think and consume zero power until they crawl back out with their answer.

2026-09-23 00:19:03 UTC

- reply

the season of wasp aggression has definitely begun. I was harassed by a couple of them on my most recent bike ride.

2026-09-08 20:00:37 UTC

A society that's so adept at fleeing their enemies, they have no need of weapons. Untouchable by virtue of flight.

2026-08-25 16:34:41 UTC

- reply

if it's plant-based fibres, the tea stain + vinegar iron rust can make a pretty rich black.

2026-06-20 17:11:11 UTC

- reply

thanks! It was a lot of work, but the work doubled as a meditation on the use of waste streams.

2026-06-20 16:45:51 UTC

I'm proud to have completed my most ambitious tablet weaving project so far. This is an intricately detailed belt with a mix of traditional and self-designed patterns. It's made from 100% cotton that I've pulled and re-spun from worn-out socks, and then dyed with recycled and plant-based materials.
It was about 2 months of work, and 5 sock-ankles-worth of cotton. I'm still working on the write-up for my website, because there's a lot to say about it.
#fibreArts #tabletWeaving #theWorkshop
https://assets.merveilles.town/media_attachments/files/116/783/394/669/629/515/original/667732aa824ef246.jpg
https://assets.merveilles.town/media_attachments/files/116/783/423/605/587/847/original/150360a0a6da5b25.jpg
https://assets.merveilles.town/media_attachments/files/116/783/424/642/831/962/original/c1d957023286005a.jpg

2026-03-23 12:55:40 UTC

- reply

ah, happy birthday! I've just turned 40 myself.

2026-02-23 18:18:19 UTC

- reply

I wonder if an easy way to calculate the contamination would be convincing enough. Like: this much plastic over this expected lifetime will cause this much local contamination.

2025-12-08 18:38:12 UTC

- reply

ah. I have a short on the stack that I'm frequently shuffling out of the way, so maybe it's a good prompt to re-examine it. I was concerned that there was something actually redundant about those three opcodes in that sequence, haha.

2025-12-08 18:24:41 UTC

I've been using uxnlin to some good effect, though I have found some confusing results:
STH2r SUB2 STH2 "Redundance"
I'll grant that I could perform the subtraction on the return stack, but I'd have to stash and swap an operand to pull that off, so it'd be the same number of instructions. Am I not seeing something?
#uxnlin #uxntal

2025-12-05 16:54:52 UTC

- reply

thousands of lambdas for everyone!

2025-12-05 16:48:04 UTC

- reply

thanks for pointing me at the fix! :)

2025-12-05 16:42:36 UTC

- reply

I had copied it while looking at it in 'less' which apparently did a tab->space conversion. Here's one copied from 'cat' without the mangled tabs: https://paste.sr.ht/~fluffysoft/167dca2db7d83ebb51d615e973ee7dc8e2399e31

2025-12-05 16:39:38 UTC

- reply

ah, sorry, I think there's some mangling of whitespace that happened in the paste. I'll see if I can get the actual contents into the paste.

2025-12-05 16:31:20 UTC

- reply

I think if you use git am to apply the patch from the paste, it'll have the right credit in the commit log, it just won't show up in the mailing list properly.
I was looking at the guide and my email provider is apparently one of the worst for this kind of thing.

2025-12-05 16:23:24 UTC

- reply

never done it before, but I've got the format right, so I think it's just a matter of getting my email server to send it?

2025-12-05 16:20:24 UTC

- reply

I checked the binary diffs and they seemed reasonable, and I ran the test script, but I suspect there isn't test coverage for this change. It's working for ubal, however. I'm guessing we'd want to update the etc/drifblim.rom.txt file if this is accepted. I have the patch pasted here:
https://paste.sr.ht/~fluffysoft/f2d7b473c3a6b4f97f8c377e3c622fec80d4d668

2025-12-05 04:29:06 UTC

- reply

Ah yeah, that wasn't so bad. My updated drifblim seems to be working in ubal, but I'll check against the tests and send a patch tomorrow, if you like.

2025-12-05 03:21:47 UTC

- reply

Are you suggesting an edit to driflim to change the lambda id generation? Seems like a good change for my style of coding.

2025-12-05 02:46:52 UTC

- reply

oh, I thought that was stated as a nesting limit. These aren't nested, just sequentially defined.

2025-12-05 02:45:21 UTC

Got a bit of uxntal+drifblim confusion right now. As I add more anonymous label sections to my code, I start getting this error near the very end:
Symbol too far: _{
Is there an absolute limit on the number of anonymous labels I can use?
#uxntal #drifblim

2025-11-23 00:19:10 UTC

- reply

ah yeah... it would take up less grid space that way. Might be extra confusing for the user, though.

2025-11-23 00:11:07 UTC

- reply

yeah, good point. Lots of kinds of values might make sense -- as long as the game+emulator knows the right place to put them. A loose grid of variables would make it easy to route that information all over.

2025-11-23 00:06:05 UTC

- reply

for absolute position, it seems like mouse is the way to go. I'll need to consider what sort of interactions will be fun to watch as well as what will be fun to edit around. I'm starting to think that the game events might manifest in Orca as bangs and zeros at fixed locations, but the "livecoder" program gradually changes melodies over time until the user starts playing with it. Once the user starts changing things, the livecoder will probably need to stop, though the events can continue.

2025-11-22 23:40:03 UTC

- reply

You're talking about the mouse cursor, right? I was thinking keyboard-based navigation might be easier to automate.

2025-11-22 23:34:38 UTC

- reply

it occurred to me that one way to have the game interact with Orca would be to have it moving the cursor around and appearing to live-code it. That'd be a delightful surprise for the player, I think :moomin_butterfly: I might need to snoop on the zero-page to read the cursor position directly.

2025-11-22 18:12:21 UTC

Sort of a question for @npub1z8r…mrq6 but also for other Orca users/modders: Would it be relatively easy to have Orca embedded in a game (godot engine) and respond to events to create dynamic bgm? For synthesis, I'm looking at Godot-CSound and Godot Synth Engine.
I'm thinking it'd be fun for players to be able to hit the tilde key and see the music being generated live and even mess around with it.
Possible obstacles I foresee include Orca's willingness to toggle between having and not having display/input authority. Also not sure if there'd be a standard way for Orca to accept the events.
#orca #godot

2025-11-21 19:39:03 UTC

- reply

thank you for saying so! I care about code readability quite a bit. The anonymous labels system is hugely useful and helps it become beautiful.

2025-11-21 17:28:39 UTC

I've been continuing work on ubal, and I'm trying to write the code in a way that helps maintain a connection to my mental model, as well as making better use of the conventions of lexing and parsing.
Ideally, I'll end up with a codebase that I (or someone else) can easily modify for use in other cultures with other time formats or numeral conventions. Even if it falls short of that, it's improving readability.
I'm fairly proud of the way I've written the lexing rules: directly in a state-machine specification. It's nested linked lists of blocks of code. You can see them here:
https://git.sr.ht/~fluffysoft/ubal/tree/main/item/src/lib/lexer.tal#L263
I've also written a method to print out all of the state transitions and the character codes on which they happen. You can find that on line 86. It makes use of nesting continuation loops, which I think makes it easier to read.
#uxntal #uxn #theWorkshop

2025-11-20 16:19:44 UTC

- reply

devilish indeed, having temptations of sweet desserts while working to stay fit.

2025-11-14 23:25:43 UTC

- reply

I've been looking at examples, and have had some chats, and did some of this stuff a few years ago, too.

2025-11-14 23:23:43 UTC

- reply

yeah, I was thinking about that when I was working on it. If I later want to do some nested scanning I might have to take a different approach.
I had an earlier version of some of these that weren't as clean with their initialization.

2025-11-14 23:09:18 UTC

- reply

I also was trying to be careful not to use any absolute addresses in the library, in case it's part of anything that gets loaded dynamically. I think I succeeded.

2025-11-14 23:04:00 UTC

- reply

ah, I thought I cleaned all those up. You found the last one, haha.

2025-11-14 22:00:13 UTC

I'm probably not the first one to do this, but I've made an interesting abstraction within the 'nint' library in ubal. The scan-le-do? and scan-be-do? subroutines scan through a large integer byte by byte (in little-endian or big-endian order) and make it easy for the caller to define operations to take place during that scanning. You can see how it's used in the addition, negation, and multiplication code.
#uxn
https://git.sr.ht/~fluffysoft/ubal/tree/main/item/src/lib/nint.tal#L114

2025-10-31 16:22:28 UTC

- reply

yeah, the "program that emits the table" is one way I was doing this before, and it's why tablerule.tal is a separate file.
I'm not sure how the values of the anonymous labels get into the table at the right positions. Is that a technique that would still require some table generating code?

2025-10-31 16:09:28 UTC

- reply

I made additional changes and pushed them. I find the tablerule.tal code a lot easier to read now, though it's probably only improved for people with syntax highlighting.
I'm curious if you have any suggestions for the rules/table section. It's a bunch of offsets, and it would be nice to have labels or some similar way of identifying at compile time that an offset is out of range for the given rule.

2025-10-31 13:50:17 UTC

- reply

ah yeah, I discovered that 'enum' style last night on your site and did some cleanup that I must not have pushed yet. Hadn't noticed the $label syntax before.
That "store where it's used" trick seems neat. I'll have to keep an eye out for opportunities to use it.
I really like the latest updates to uxntal, especially the nice new syntax for calling subroutines ( a bit more FORTH-like ). It's fun to use drifblim for compiling, too.

2025-10-30 21:44:51 UTC

I've gotten back into writing 'ubal' which is attempting to provide some plaintext accounting tools for the uxn varvara ecosystem.
I was inspired by a recent bookkeeping workshop that I gave. I like the plaintext accounting tools that I use, but I believe they could be a LOT smaller. It's also an excuse to imagine other ways things can work.
I'm quite pleased with the way I've implemented the parser. Still a long way to go with feature implementation, but it's feeling like a solid foundation.
https://git.sr.ht/~fluffysoft/ubal
#uxn #theWorkshop