iiinova — where your digital self keeps evolving

NON-TOXICNo ads · No shady algorithms

iiinova is part of the Atmosphere — an open, public network where you own your identity.

mike bayer :python: :redhat:

@zzzeek.fosstodon.org.ap.brid.gy

15 followers 2 following 95 posts

cats, dogs, python. drum-related content at @zzzeek 🌉 bridged from ⁂ fosstodon.org/@zzzeek, follow @ap.brid.gy to interact

Posts Replies Media

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 04/11/2026

Just to sum up yesterday: 1. as soon as you put one file on pypi from CI with attestations, you now have to use CI releases / attestations for *all files, forever*, or you will get large red warnings on your package (and they dont send you emails either). 2. if you dont use attestations at all […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 03/11/2026

so a little clarity on what's happening here. If I *never* used trusted publishing *at all*, I'd have zero of these banners all over my page. pypi is throwing all this red paint on me because I use trusted publishing for *most of my wheels*, as I was encouraged to do. but because my source .tar […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 03/11/2026

do folks remember the discussions here about "attestations"? so many people told me I was misunderstanding "attestations", how they had nothing to do with your existing workflow from your workstation? look at pypi now. Clearly they are trying to SCARE THE […] [Original post on fosstodon.org]

the pypi page for SQLAlchemy 2.1.6. With a giant red banner that says "
Incomplete provenance

8 of 72 files were uploaded without provenance metadata. View details. " complete with triangle exclaimation point, and then a giant red box down below that explicitly says: "We recommend further investigation before installing files from this release. Learn more. "

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 03/11/2026

this is apparently correct - Google is refusing to provide a google search interface that does not force the "AI summaries" into the top. that's a very shitty decision because it's pure marketing and marketing has no place in kids education (even though it's sadly ubiquitous)

Google Will Turn Off Search for Schools That Reject A.I. https://www.nytimes.com/2026/11/03/us/google-search-schools-restrict-ai.html?utm_source=flipboard&utm_medium=activitypub Posted into U.S. @u-s-newyorktimes

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 02/11/2026

SQLAlchemy gets relentless bug reports now. Many are regressions from the recent 2.1, that's normal, but lots from 2.0.x too. Most of these are real issues that affect real apps; the frivolous "AI fuzzer" ones have declined a lot, as the community has pushed back on those. What is not normal […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 02/11/2026

This is a really well written article about AI and interestingly it's the very predictable reader comments that all sound like they were written by an anti-ai-talking-point robot that is depressing www.nytimes.com/2026/10/30/opinion/…

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 28/10/2026

if LLMs manage to find working treatments for MD, every "using an LLM is unethical" argument is gone. this is a very tragic and horrible disease. if we in fact built a machine that can find a cure for a disease like this, it would be a truly amazing achievement

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 28/10/2026

here I was today thinking, that PHP thing was kind of dumb, because first of all, PHP, second of all, it was a one off that's possibly not even practically exploitable , people are freaking out a little too much about the overall "oh no endless 9/10 vulnerabilities for years!" but here we are […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 21/10/2026

I have to patch some servers that I maintain on a familial basis but I am not calling myself a "PHP user" when I do so, sorry

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 21/10/2026

happy "shut off all your php sites until 10:30 am" day

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 20/10/2026

I cannot believe Google won't give us a control to mark and block multiple messages at once as spam, or even a control to auto block these. It detects them all as political spam identically but refuses to not send them due to some stupid legal loophole

Screenshot of nine political text message threads in a row in one day (only a partial listing)

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 16/10/2026

some serious unicode fail today

screenshot of an activation email:

You‚Äôre almost there!

Click below to activate your account.

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 15/10/2026

WE ARE GETTING THIS CHESS ROBOT!!! but I have to wait for my son's birthday to play with it!! www.youtube.com/watch?v=fha3WC-3DOg

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 13/10/2026

AI companies read all the documentation of open source projects without permission and repackage it for profit - without paying the maintainers!

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 12/10/2026

LLM opponents, a question. how do you folks google for "how do I do X in Python", the way we would in the old days and get directed to StackOverflow, and not use AI? Google at least will put front and center a code example its LLM just worked up which is usually correct. Then you can drill down […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 12/10/2026

i just tried Codex for about ten minutes, seeing if it could run through some of the basic workflows I've used with Claude, and it did so terribly I ran screaming. it's funny how all these models have similar capabilities but Claude has really tuned it for "programmer working with code"

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 09/10/2026

well my genius mind still annoyed by the constant onslaught of "AI does not improve productivity" studies being dangled around as some kind of indictment of LLMs, thought wait a minute. I bet they said the same thing about word processors (and I have used actual typewriters, for pay, in an […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 04/10/2026

I dont know much about Yan LeCun except I saw a well known activist blaming him for the Rohingya genocide, so I was like OK.."works for facebook" is bad. but he's pretty on point in this article […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 04/10/2026

today on bluesky, four studies showing interactions with correctly-aligned LLMs can be used to dispel conspiratorial thinking, even when perceived as human pubmed.ncbi.nlm.nih.gov/41179846

pubmed.ncbi.nlm.nih.gov

pubmed.ncbi.nlm.nih.gov

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 02/10/2026

after today's events I'm probably going to be a little more rude/blunt to hardcore anti-AI absolutists.

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 02/10/2026

im thinking more and more about how I was affected by the introduction of drum machines as a common way to replace the need for drummers in the 1980s, just as I was hoping to start a career as a drummer. so far it at least seems to give me the "this machine can do XYZ better than any human and […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 02/10/2026

so the inevitable task we finally did in SQLAlchemy 2.1 of changing the deafult DBAPI driver for postgresql:// and oracle:// URLs (because DBAPI drivers change in prominence too!) is hitting a few surprised users enough that they're breaking out the semver argument (URL stuff is where we […]

Original post on fosstodon.org

Reposted by mike bayer :python: :redhat:

Glyph @glyph.im · 01/10/2026

Taylor Lorenz is preparing for her "why I left the left" right-wing grifter pivot. Do not trust her. www.thefp.com/p/two-drinks-with-tay… (If you do not know "the free press", it is Bari Weiss's right-wing false-balance propaganda outlet, so talking to them *at all* is […]

mastodon.social

Original post on mastodon.social

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 27/09/2026

we're on hour 40 or so of this noreaster thing. non stop winds, being outside for any amount of time is inherently dangerous. All of Long Island is like hundreds of outages all over the place, not from the usual hurricane fast destruction thing, just by persistent winds hammering the trees hour […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 27/09/2026

"How to keep enjoying programming in a world of LLMs" mostly good advice which is close to how I work with agents as well. For me it's very much an "I review the model, the model reviews me, then another model reviews both of us" kind of thing […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 25/09/2026

OK so far I am not liking Control:Resonant as much as Control. it's much heavier fighting right from the get-go and it's much more button-mash style, it's physically exhausting. being able to throw objects in Control felt amazing and so far none of the stuff I can do feels like that, it feels […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 24/09/2026

next on deck: * release SQLAlchemy 2.1.0 (it's been three years since a major release) * play control:resonant and drink wine all weekend

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 18/09/2026

my most requested feature on github now is a new markup style that can denote LLM content in a comment or post. I'm finding my general method of communication is write a "cover letter", hey im Mike, im a human, here's my thing and the problem I'm solving, and I used Claude to help with it, then […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 18/09/2026

so the other day a DIY bird-identifier frame project got 2297 points on Hacker News. Refreshing in its simplicity, aesthetics and decided emphasis on "human made", hundreds of responses from hackers what a great idea, and i should do this for my mom / spouse / whatever I am here to tell you […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 15/09/2026

this is definitely what freaks me out the most about LLMs I have to pick and choose all day what I feel I need to understand and what I can discard. this is not an easy process and my current thinking is that this is a new skill everyone involved in knowledge work will have to master. it's […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 15/09/2026

no, you can't prevent "AI" by banning datacenters and taxing billionaires. "AI" is being trained around the world. Models from China are nearly as good as US models. Open weight models are already available everywhere. Models that can coordinate hacking […] [Original post on fosstodon.org]

And if one of our LLMs in a massive hyperscale AI data centre somehow magically became sentient, it's not like we could just shut off the water. Or the electricity. Or stop delivering massive quantities of gas and oil to the onsite gas turbines.

Or just find the nearest Telstra pit and put a bloody shovel through the fibre optic cables that connect it to the outside world.

So we need to immediately figure out ways to stop this happening!

What's that?

You say we could prevent this happening by just not building the massive AI data centres?

You say we could tax the rich so they don't have the money to build massive AI data centres?

No! We absolutely must not do that under any circumstances!

Reposted by mike bayer :python: :redhat:

Nicholas Grossman @nicholasgrossman.bsky.social · 15/09/2026

AI psychosis and anthropomorphizing chatbots are a bigger problem than the risk of spontaneous emergence of super-intelligent consciousness. LLMs can already do a lot, will be able to do more, and are changing society. Yet much of the worry assumes they have dreams, desires, independent thoughts.

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 14/09/2026

one of the most widely circulated studies showing "early 2025 AI slows down experienced open source developers" has a prominent badge pointing to the new version of the study, which reverses the findings given later improvements in agentic coding. New issues in […] [Original post on fosstodon.org]

from https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/#motivation

These results are out of date.

We have released results that are current as of early 2026, in a continuation of this study. We believe these historical results no longer reflect the current impact of AI models on open-source developer productivity.

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 11/09/2026

this is the core thing we should really be talking about there's no debate that Anthropic/OpenAI are fanatical companies who need a lot more oversight "The issues the mathematical community faces now are similar to issues that other scientific and creative professions are facing, and indicate […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 11/09/2026

"It’s a good example of how coding bots are a legitimately useful tool for developers. It looks like he didn’t use the bot to write the fix – that wasn’t needed, it’s a one-line fix – but the bot wrote 42 chunks of test code that enabled Linus to find the bug. It sounds like that was a tough job […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 11/09/2026

here's a prompt i just did just now. grunt work. happens constantly. some DBAPI puts out a new version that fixes one of their bugs; one of our "expected fails" for them breaks other builds in our CI. this is not "project initialization", that is not the only […] [Original post on fosstodon.org]

a prompt written in the vscode claude plugin:

it looks likje perhaps a new version of mssql-python has come out and it's causing CI failures in the form of unexpected success: https://jenkins-static.sqlalchemy.org/static-files/RaOG_jFmaUJNoj-CEhZ5dC8JaRK1-eMaNkjSfigrbuwxNzg5MTQzNTAwMDQ2Ojc6Y2xhc3NpYzpqb2Ivc3FsYWxjaGVteS9qb2IvZ2Vycml0XzIxLzExNDkvYXJ0aWZhY3Q=/FAILED-py314-mssql-cext-backendonly-test-report.html can you evaulate and do an unticketed gerrit for main/2.0.x that adjusts this feature for whatever changes in mssql-python may have just been released?

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 11/09/2026

i'm still really super all in on gerrit for code review. when i have to do full reviews on PRs alone I am sad. My whole flow with claude or whatever is, yeah do what youre doing and then get it on gerrit. that's where I read what it's doing. I make my comments there and have it push up a new […]

Original post on fosstodon.org

mike bayer :python: :redhat: @zzzeek.fosstodon.org.ap.brid.gy · 10/09/2026

hi folks, back on fosstodon. no reason