Computers are bad 🏳️‍🌈

Teaches maths/computer science at The University of Leeds

Leeds, UK

Pronouns: he/him

Last Notes

- reply

Can I redeem my achievements for monetary and/or spiritual rewards?
Get snacpilled, Emperor of Cabbages, Inner circle (iirc), Godkiller, Look ma' I'm a webdev!, S T R O N G C O N T E M P T (if mothra counts?), Choose life
I think there should be one for migrating to snac, like "Manifest Destiny" or something less/more evil

2026-08-28 08:06:32 UTC

(Thinking of text editors) someone should set up a PO Box that you can send handwritten/typewriter'd code to, which is then compiled and run, and the results sent back. Like in the olden days

2026-08-23 17:08:48 UTC

9front, featuring snac, a German ThinkPad R40, a toy accordion, and other assorted tat
Took a few attempts to burn a DVD that would boot, might be either a USB DVD drive issue or an OpenBSD issue (as it worked when I burned using Debian with a built-in drive)
The BIOS mentions booting from USB, but I think it means booting from an external USB DVD/floppy drive (not a USB stick). It also seems to have wifi, but OpenBSD doesn't have the drivers so I assume neither does 9front
https://patpatpat.xyz/snac/pat/s/post-a45940b38cef0ccff71e955f240f625a.jpg

2026-08-21 18:53:55 UTC

- reply

Hehehe, this may be on the podium for the worst way to calculate factorials:fact = (u . R (1+ zero) (u v . (x1 x2 . * ((y1 y2 . 1+ (1/2 y1 y2)) x1 x2) (2/2 x1 x2)) u v) u)<br>
Where 1+ is the successor function, zero is in Church numerals, * is defined in the usual way, 1/2 and 2/2 are projections (i.e., n/m returns the nth argument of an m-ary function)
Note: it takes over a minute to do 4! the way God intended
Note 2: I didn't just come up with this by hand, Hindley and Seldin's "Lambda Calculus and Combinators" describe a recursive process to convert primitive recursive functions into lambda calculus terms (but I did have to do the conversion using my hefty noggin)
Note 3: Can be simplified slightly:fact = (u . R one (u v . (x1 x2 . * ((y1 y2 . 1+ y1) x1 x2) x2) u v) u)<br>

2026-08-21 17:25:48 UTC

Aha, I *think* I may have fixed my reduction problem with https://patpatpat.xyz/data/lambda/
It was getting stuck when attempting to use the Y combinator. In particular, you can derive the R combinator as a fixed-point using Y, which should still satisfy R A B 0 = A and R A B n = B n-1 (R A B n-1)
Before it was recursively reducing every sub-term until each one reached normal form, but now only the top term is reduced until normal form
Not quite sure why this worked (perhaps I was assuming something about how child reductions affect the parent?), but it seems to work with my primitive.lambda file (which implements primitive recursive functions in lambda calculus)

2026-08-20 18:32:51 UTC

Love #snac
I kept fat-thumbing when trying to open content warning'd/hidden posts on me phone...but I realised I can just edit the style.css to add a bigger margin. I should probably check out some of the proper stylesheets people have made (but there is something very rad about the default theme)
https://patpatpat.xyz/snac/pat/s/post-c902a3bec55ba4fe780e4bc959351169.png

2026-08-20 13:16:09 UTC

- reply

That is a highly believable conspiracy. Ah I've not encountered that pop-up yet (I almost never look on Reddit), I assume it is waiting for the perfect moment to strike...

2026-08-20 10:55:03 UTC

- reply

This annoyed me. I've been using the Old Reddit Redirect extension since forever, so it was a bit of a shock to see how utterly awful "New" Reddit is. It makes random words into hyperlinks, but they just link to the shitty AI thing. Why?!?

2026-08-07 09:23:06 UTC

I got my degree from a ballerina school, it was a 2:2

2026-07-13 15:37:39 UTC

Proof left as a punishment for the reader...

2026-04-25 17:28:30 UTC

- reply

I've definitely read books that I've hated, and immediately forgotten. So at least its not one of those haha
I really like that sort of ye-olde, matter-of-fact style. Its direct and efficient. I'd like to be able to write like that (my poetry is, but my prose gets too flowery/verbose)
Is the Monomoke one like Beren & Luthien? (I need to watch Monomoke again!)

2026-04-25 16:08:33 UTC

The Silmarillion is fantastic, surprisingly readable too (there are quite a few characters, but each story only involves a few key ones). The geography is the hard bit imo
I may even have enjoyed it more than The Hobbit and LOTR (don't quote me on that)
I've built up this mental image of Beleriand, which is utterly useless (i.e., has no relevance to other Tolkien stories)
The way all the histories/lineages/geographies compound on eachother is incredible. Everything echoes, and those echoes are embodied in the societies and physical landscapes

2026-04-14 16:06:59 UTC

- reply

I love these combinator names, next step is to draw cute pictures of each one ;)
(You've probably seen this already, but...) these lambda calculus ones are especially cute: https://dkeenan.com/Lambda/

2026-04-08 16:36:16 UTC

- reply

Ah yeah that makes sense, it feels a lot more intuitive that way
Ooh I didn't even think about optimisations. Do you mean that x^10 'y/x would become y^10/x^10? (i.e., from an anonymous label draining x, to a fraction producing y^10)
Its crazy that "optimising a program" could just be "simplifying fractions"

2026-04-07 23:01:44 UTC

- reply

hehehe, a text-based adventure game is theoretically possible. I added a naive "?" command to get a line of input (its then split into words, and those words are put into the bag)
I think user input will be fiddly though, because its essentially giving the user a REPL with unlimited access to the program. Perhaps limiting the input to a specific subset of words would work, like ?[look take quit]
https://patpatpat.xyz/snac/pat/s/post-cef560941cf2e8f5491e854486dda198.png
https://patpatpat.xyz/snac/pat/s/post-da45bbae4ff0f6781f8e0a6a8a4cb6ce.png

2026-04-07 20:33:42 UTC

I implemented @npub1z8r…mrq6 's Rejoice, in Perl for some reason :)
This was attempt like 3? I vaguely remember reading a Moore quote about "writing, rewriting, and rewriting again". I follow that advice religiously
I absolutely hate writing parsers, and yet invariably end up having to write a parser. This is probably the best one I've done but its still a bit of a mess!
The only difference I can spot between my implementation and the UXN one is that mine can't differentiate between [x x x] and x^3 (it always uses the latter). This means that things like [ .x .x .x ] will only print a single x (I'm not sure what the intended behaviour is for that case anyway tbh?)
https://patpatpat.xyz/data/rejoice.pl/

2026-03-22 18:29:41 UTC

- reply

It seems like there is an element of coding-style involved too? A lot of the FORTH/Uxn code I've seen splits lines up quite deliberately, which I imagine helps the author understand what's going on
There's probably no "one-size-fits-all" style guide though?

2026-03-22 17:57:22 UTC

I assume at some point I'll be able to write FORTH without stack annotations after every instruction, but its not today...
(Reminds me of how tricky it was to get used to parentheses in Lisps)