iiinova — where your digital self keeps evolving

rkletr @merkletr.ee · 29/09/2026

i don't actually wish for this to be clear. i do hope that machines can eliminate all work and lead us to post-scarcity, but until then... i have fun with writing code and it's a point of pride that i'm still better at it than the machine that's probably gonna take my job in a few years.

010

rkletr @merkletr.ee · 29/09/2026

i didn't think it was core to my identity that i know sql but apparently it is

001

rkletr @merkletr.ee · 29/09/2026

i wish i could delegate all my programming to the machine god but the machine's capabilities are jagged and unfortunately lag behind in areas i consider critical

110

rkletr @merkletr.ee · 29/09/2026

this post brought to you by llms once again being terrible at software engineering in ways that to me seem egregious until i once again remember that the average software engineer actually sucks at software engineering and llms are probably above median

010

rkletr @merkletr.ee · 29/09/2026

designing normal databases has probably been one of the more important things a cs degree taught me (in my case moreso gave me vocabulary and formal framework for what i already knew intuitively but still) if you don't intuitively get sql schemas i would highly recommend putting the effort in for it

110

rkletr @merkletr.ee · 18/09/2026

it's wild how common it is for people to think "large systems just always have vulns" it is quite hard to make software with zero bugs, but it's not impossible most large systems just have lots of vulns because people do not care enough about systems they build to actually rigorously engineer them

080

rkletr @merkletr.ee · 18/09/2026

man am i gonna have to make my own software engineering benchmark just cause you can't trust any of the mainstream ones

020

rkletr @merkletr.ee · 18/09/2026

maybe the reason llm benchmarks mostly suck and dont represent performance is just cause the average llm user cant stop themselves from making slop how hard can it be to just actually manually go through and audit your damn benchmark tasks before releasing it to the world

130

rkletr @merkletr.ee · 09/09/2026

in the weird part of the night that's the time i feel alright

000

rkletr @merkletr.ee · 06/09/2026

ive only lived in quebec haven't done tourism in it... i should try that at some point

030

rkletr @merkletr.ee · 06/09/2026

living in montreal and doing tourism there are such different experiences (and i love both)

160

rkletr @merkletr.ee · 06/09/2026

wish i could tell transit to always show me metro-only trips first even if they're slower. i *will* choose a slightly longer trip if it means avoiding stm buses

000

rkletr @merkletr.ee · 04/09/2026

forgot about this i should go through my wishlist and get some stuff

010

rkletr @merkletr.ee · 03/09/2026

idk we'll see i do plan to consider other approaches entirely

010

rkletr @merkletr.ee · 03/09/2026

after raytracer (3kuymtsx53k2y, 3lm656ah3622k) i thought it would be fun to do rasterizer i don't think i'm gonna be able to fit this in 300 after adding shading, especially since some of the tricks i used to golf this wip are evil (e.g. got rid of some opts, renamed the input from teapot.obj to i)

110

rkletr @merkletr.ee · 03/09/2026

i:(1_'i)@=`s$*'i:" "\'(97>*:')_0:`i w:640;s:w,h:480 v:@[+.''i`v;1;-:] p:0.5*s+h*(2#v)*\:q:1%1-0.2*v@2 f:((.*"/"\)''i`f)-1 d:{-/x*|y} (a;b;c):++'p[;f] D:1%d[b-a]c-a "o.ppm"1:("P6\n640 480\n255\n"),,//{{`c$$[#k:&~0>&/e:(1-B+C;B:D*d[x-a]c-a;C:D*d[b-a]x-a);_255*e[;k@*>Q:+/e[;k]*+q@f@k];&3]}'(!w),\:x}'!h

120

rkletr @merkletr.ee · 02/09/2026

explain a vaguepost? why would i ever do that.

110

rkletr @merkletr.ee · 02/09/2026

i should stop using this account and start using only the other one

220

rkletr @merkletr.ee · 01/09/2026

oh shit turns out clang-format works. sure i'll use clang-format for typescript why not

---
BasedOnStyle: Microsoft
IndentWidth: 4
---
Language: JavaScript
AllowShortBlocksOnASingleLine: Always
AllowShortFunctionsOnASingleLine: Inline
AllowShortIfStatementsOnASingleLine: WithoutElse
AllowShortLoopsOnASingleLine: true
BreakBeforeBraces: Attach
Cpp11BracedListStyle: Block
InsertNewlineAtEOF: true
JavaScriptWrapImports: false
SpacesInContainerLiterals: false
UseTab: ForIndentation
...

040

rkletr @merkletr.ee · 01/09/2026

wtf is there not a single _un_opinionated js formatter. why is the whole industry just copying one other formatter's defaults

341

rkletr @merkletr.ee · 30/08/2026

(lol) in more seriousness my harness might not be what you want at all rn but i _am_ planning some stuff for per-project plugins so idk maybe in a few weeks it will be just gotta unslop the current codebase first

110

rkletr @merkletr.ee · 30/08/2026

that's kind of my harness' whole deal!

100

rkletr @merkletr.ee · 21/08/2026

oh no deepseek has vision now that means i finally need to go and actually implement image support into my harness

040

rkletr @merkletr.ee · 19/08/2026

hmm maybe not great for atproto which uses them for basically everything (services like the bsky appview or feed gens or moderation services, etc get the key from the user's did in iss, with aud+lxm being the service+route)

010

rkletr @merkletr.ee · 14/08/2026

im reading the paper soon i think they might make up on the complexity by being real fucking nerds (compliment) about it

010

rkletr @merkletr.ee · 14/08/2026

eh it's definitely more generic and flexible than tame, maybe that's worth the complexity to them. there are a whole lot of things (e.g. streaming) that i don't care and i can afford to not care because my harness is not for the general public. which like ofc that changes the tradeoffs i make

100

rkletr @merkletr.ee · 14/08/2026

hmm idk how much of a fan i am of deepseek harness cordis

100

rkletr @merkletr.ee · 13/08/2026

sorry; there was a power outage at home again, should be back up now (my PDS runs on the same server as the feed so both are down usually at the same time & so i can't log into this account to announce when the feed is down)

100

rkletr @merkletr.ee · 11/08/2026

i am now finding out this is actually like a super common thing and there are a ton of alternatives including multiple names i fucking recognize but just never actually thought of for long enough i guess

000

rkletr @merkletr.ee · 11/08/2026

also the interface is a buggy slow ugly hard-to-navigate mess

100

rkletr @merkletr.ee · 11/08/2026

like, it goes and traces data flow through your code and tells you about potentially unexpected vulnerabilities?? and it runs decently fast so you can put it in CI?? too bad the queries are vendor-provided instead of like in our git and they suck so bad that all of its findings are false positives

100

rkletr @merkletr.ee · 11/08/2026

this software we use at work (CxSAST) is actually a really cool concept tbh i wish it wasnt so damn mediocre and painfully annoying

100

Reposted by rkletr

rkletr @merkletr.ee · 06/08/2026

this is a video of me pressing ctrl-backspace in jetbrains rider

121

rkletr @merkletr.ee · 06/08/2026

this is a video of me pressing ctrl-backspace in jetbrains rider

121

rkletr @merkletr.ee · 05/08/2026

every "smart" feature usually just breaks years of accumulated muscle memory and forces you to learn a context switch between an IDE and every other text editing interface full-line+ completions, VS inserting indents when you navigate through empty lines, per-language ctrl-backspace behavior, etc

130

rkletr @merkletr.ee · 05/08/2026

the more the text editing experience in an IDE diverges from text editors the worse they get

130

rkletr @merkletr.ee · 05/08/2026

man i hate feature flags so much

010

rkletr @merkletr.ee · 04/08/2026

unless they've banned me or fucked up their ingest and by "can't see the feed" you mean the feed literally doesn't exist there

101

rkletr @merkletr.ee · 04/08/2026

hmm i don't see how that could happen, assuming the feed works from the same account when you're using the pbc appview. the pds constructs the jwt for the feedgen, and if the appview fucked up forwarding it with getFeed you presumably wouldn't be getting _any_ personalized feeds...

100

rkletr @merkletr.ee · 04/08/2026

ouai ouai is kinda peak tbh it was my least preferred of the trilogy at the start but it's really grown on me

000

rkletr @merkletr.ee · 03/08/2026

i dont keep request metrics but i wouldnt be surprised lol there are only 129 people who even followed someone with the feed, let alone actually use it

grafana stats for priv: posts, follows and users, all-time and from the last 24 hours

110

rkletr @merkletr.ee · 03/08/2026

have a grafana alert, should no longer be a problem in the future (or at least i'll know when it happens)

100

rkletr @merkletr.ee · 03/08/2026

fuck me i hadnt noticed. restarted it now but i don't know exactly from how long ago it'll be able to backfill, we have have lost a few days there

210

rkletr @merkletr.ee · 03/08/2026

huh the stm doesnt have real-time metro data. thats annoying

110

rkletr @merkletr.ee · 29/07/2026

idk your code is probably fine lol im just complaining about work

010

rkletr @merkletr.ee · 29/07/2026

probably because all go code is bad code so you can at least consistently expect code which is tolerable level of garbage whereas C# code is forged into bad code through the fires of Entreprise; venturing into C# codebases is equivalent to throwing yourself into pits deeper than you can imagine

130