pleroma.debian.social

pleroma.debian.social

Are.

You.

Fucking.

KIDDING.

Me?

https://lists.debian.org/debian-vote/2026/08/msg00197.html

In the midst of a heavy discussion and voting about whether this should be at all allowed…

one arsehole, with a history of shoving slop into people’s faces and shoving people’s mails into slop, decides to present a vommit ostensibly fixing an issue in the program used to tally Debian project voting, …

… AND THE PROJECT SECRETARY, WHO RUNS THE VOTING, SAYS HE WILL JUST FUCKING MERGE IT‽

šŸ™€šŸ¤Æ

I just…

I just CANNOT.

This is in-your-face. This is beyond pale. This is multiple insults in one. Even if this were legit, it would be in utterly bad taste at this precise point in #Debian project history.

Let’s rename to Slopbian, will you?

oh for fuck’s sake, jrtc29 pointed it out, and the slop arsehole just said he is ā€œinterested in closing this topic as soon as possibleā€ and meanwhile ā€œdo[es]n't see why [he] should respect a rule that does not exist yetā€

I absolutely cannot even.

the slop user cannot possibly stop responding, it’s a trainwreck in the pipermail

mhonarc, whatever, thankfully not that mailman3 thingy

why is devotee (Debian vote engine) not packaged?

the delusions of some of those slop advocates…

It's not a "better autocomplete". It is a pattern matcher. It looks at a text and detects patterns in it - double linked list here, factory pattern there, singleton here, regex for an IP address here, ... And then when generating text it can apply all the patterns it knows

ā€œit knowsā€ 😹

Someone clearly doesn’t know how LLMs work.

there’s just so much wrong in both the underbelly and in areas I rarely have reason to engage with, as I’m normally just quietly doing my thing and using the same software I did decades ago plusminus very few bits…

@mirabilos Sometimes makes me wonder which parts of Debian infra are actually using packages.

@lanodan the buildds are, nowadays, but, yeah

@mirabilos What, writing stuff (& coding) is not primarily about pattern-matching with thought-terminating clichƩs?

@tkchia LLMs cannot recognise patterns (as claimed by AIgars), is what I wanted to say

@mirabilos yeah, this is why am I rather exhausted from that discussion. Granted, I did start it, but how detached from reality the LLM proponents are surprisingly even me.

@werdahias yeah, when people, including among their numbers some whom you liked, trusted, worked with before, who you would have thought to know better, both hand-before-the-eyes and deliberately are setting new fires to FOSS and other areas touching you and your life while you’re busy trying to put them out and trying to figure out how to save what’s left, while at the same time quite literally setting fires to the planet (we’ve had hundreds of hectares of forest fires worryingly close to ::1, one of the fires alone having been larger than all forest fires of the state of the last eight years together), then…

… then you’re entering fatigue.

@mirabilos yeah, and they don't even realize it. It's rather tiring to the point where I already decided to contribute more to the KiCad libraries for now because they have a strict ban on slop.
replies
0
announces
0
likes
1

@mirabilos True that. LLMs are based on fuzzy probabilities rather than clear-cut patterns.

@mirabilos Some things I've observed when playing around with GitHub Copilot (the free offering) on my own code:
- it's pretty good at "understanding" things such as, say, "x *= A" means to multiply x by the constant A;
- if it seems to "understand" the high-level reason I did something a certain way, it's because I explicitly wrote it down somewhere;
- it keeps trying to rewrite a non-GitHub project URL to point into GitHub (wot)...

@tkchia no, it physically literally cannot understand