iiinova — where your digital self keeps evolving

rini.tngl.sh @rini.tngl.sh · 21h

i let gadts get to my brain too much this IS a regular adt

000

rini.tngl.sh @rini.tngl.sh · 22h

should i stream haskell

150

well its awful without a GADT too i guess but i could at least have a map to the argument types like in the typescript version

maybe, because i dont know how to implement `Get` on it besides also generating something like this

i usually disable the client-side image compression bc its bad and also i need to give bluesky canvas access too late every time

000

will I regret using a GADT like this (auto-generating this code though)

230

noo dont like my resistFingerprinting fail again..

000

omg discord tree-shitter

why he flashing that jerma smile

rini.tngl.sh @rini.tngl.sh · 05/10/2026

til -fdefer-type-errors haskell has a typescript mode

060

oh this might actually be usable. myabe

000

being meguca is suffering

060

cloudflare made open weights jev???

000

bouncing between "holy shit this is epic" and "no i actually hate types"

1120

Glasgow "No we don't have dependent types" Haskell Compiler

2151

care to explain why Nat, "not-a-type" is actuavly a type?

rini.tngl.sh @rini.tngl.sh · 03/10/2026

genuinely deranged that they recommend u use a random flake that makes this accessible as an attribute list bc thats all flake refs support

rini.tngl.sh @rini.tngl.sh · 03/10/2026

genuinely how do u do the former cleanly on nix3

rini.tngl.sh @rini.tngl.sh · 02/10/2026

is there no System.IO that uses System.OsPath..?

000

rini.tngl.sh @rini.tngl.sh · 02/10/2026

the next line starts with Q

000

rini.tngl.sh @rini.tngl.sh · 02/10/2026

to me everything jj fixes can be done with commit --fixup and rebase --autosquash

not saying who but someone's ssh private key got leaked, you might wanna check yours if it starts with `b3BlbnNzaC1rZXktdjEAAAAABG5vbmUAAAAEbm9uZQAAAAAAAAABAAAAMwAAAAtzc2gtZW` its compromised, i have it saved ♡

2163

26.11pre1078696.4975466d3247 -> 26.11.20261001.c59305b nixos-unstable version format change? but the tarball url for this still has `26.11pre1083223.c59305bab206`

it hates me and refetched everythinb

git show HEAD^:npins/sources.json | jq '{ nodes: .pins | (.[] |= { locked: if .type == "Channel" then { narHash: .hash, rev: "", type: "tarball", url } else { narHash: .hash, rev: .revision, url } end, original: {} }), root: "root", version: 7 }' >flake.lock

I read someone write "actually, its humans that should be using little-endian instead of big" ages ago and still think of it

070

eat them, with a little garlic though

000

I wonder if by pure chance there already are sha-256 collisions out there, just in different systems so they do not know about each other...

160

..possible, but nothing works currently

also i didnt kill the webpack code, so if witchsky stays on webpack it should be possible to write plugins for it too although i dont know anyone who would

130

add "literally replaced webpack with metro" to the list of things that Did Not break my regex patches, for all my non-believers

140

oh okay i fixed my bluesky client mod already?

rini.tngl.sh @rini.tngl.sh · 29/09/2026

>rm pnpm-* >npm i >immediately notice that i somehow have eslint@9 using eslint/js@10 and pnpm did not fail

rini.tngl.sh @rini.tngl.sh · 29/09/2026

why does every GC (RAII is a static GC) delete the object on scope end instead of last use

rini.tngl.sh @rini.tngl.sh · 29/09/2026

graph theory tonite? 👀

so sorry for the two users of my client mod (me, my friend) it wont be fixed until tomorrow

absolutely deranged but it mostly works the same so whatever

what. bluesky uses metro on web instead of webpack now???

oh guess my client mod died hold on

000