sebazzz

Intent is hard to prove if it is not published.

Yup, Windows Azure App Service is just IIS.

When it comes to the big AI players feeding their future models I 100% just assume that they suck up any data we send them. Am I cynical?

There is a reason enterprise contracts and plans exist. And I think even on that account we're going to find out at some point that LLMs are training on that extremely useful data.

What I fail to understand: Who or how are these AI generated pull requests paid for? There is nothing to gain financially by creating all these AI generated pull requests - yet the increasingly more expensive tokens have to be paid for somehow. How is it paid for? Or are these authors using all their free GPT5.4-mini tokens from Github until hitting the rate limit, day after day? Or are they running some local AI model on their GPU like Qwen and heating the room in they work in, day after day?

How is the stability? A few years ago I could barely run The Great Green Mouse Hotel because of emulation flaws causing the CPU to do weird things and either crash apps or the Windows 98 itself, or the emulator.

That is possible, but if swapping happens due to memory pressure it can still cause slowdowns in unrelated parts of the same or other processes.

If Bambu Lab goes after Orcaslicer, then it makes sense they will get scrutinised for the AGPL violation.

P.S. You should play a lottery/powerball ticket

Actually, they should not. That collision and winning the lottery would be even rarer.

Well it would be statistically even rarer for that UUID collision to happen and the earth to be destroyed by an asteroid.

What is the difference between this and using normal OCR and then running that output through a LLM? It seems such a bazooka way to kill a fly to me using a modelime Qwen.

Yes it is very good cinematic. Unfortunately it is far from the truth.

Haiku, right?

Who would have put "Trump accelerates the green energy transition in 2026" on their bingo card?

Even with all the simulators and months of rigorous testing, sending up something that can (in the worst case) break the probe has to be terrifying.

I would guess that even that case is partially accounted for by a watchdog that is hardwired into the system.

It is quite embarrassing to ship a CPU emulator with broken floating point math as the default.

It was also quite embarrassing to ship a real CPU with broken floating point math.

With UnsafeAccessor you can often avoid reflection.

I wonder how the voting components are protected from integrity failures?

They are quite expensive and there is not something similar on the market (even not from house brands of Aldi, Lidl, etc).

You still have things like git squash etc.

Ouch, reminds me of hours debugging OAuth2 implementation in my Surface 1 app for Twitter because the nonce or some other checksum was not calculated correctly.

If I am not incorrect, for early boot applications, the application must be set to use the native subsystem.

Apparently they are recruiting from Kenia too, promising great pay, but in reality they are being abused as cannon fodder.

Also take the randomness out of it. Sometimes the agent executing tests one way, sometimes the other way.

> so you need to tell them the specifics That is the entire point, right?

Honestly it is a problem with using GPT as a coding agent. It would literally rewrite the language runtime to make a bad formula or specification work.

That's what I like with Factory.ai droid: making the spec with one agent and implementing it with another agent.

US software providers can shut down their software services in an instant, paralyzing European societies.

And no US SaaS provider or cloud provider will ever be trusted again, instant cutting off a part of the US economy.

Meanwhile for weeks at a time I got obvious spam and phising right in my inbox. It just didn't stop. Twenty mails per day. I was about to cancel my Office 365 subscription; and then from one day to the other it just stopped.

it turned out that the version in source control wasn't the version running in production and it would have been quite a lot of work to reverse engineer the production version

When I started at my work, a previous software dev with practices more like a mechanic than a software dev didn't use tags and all binaries deployed to production were always the default version 1.0.0.0 of the C# project templates in Visual Studio. To make matters worse, variants of the software were just copy pasted in CVS with their core code checked in as binaries and not their original C# projects. Fun times finding out what actually ran on production, and patching anything in it!

It is just a button. Default it starts the Copilot app which is really the Office app that already existed but now with the copilot tab preselected. Also that Copilot runs in the cloud and doesn't use your NPU.

The only thing until now I've found using the NPU are the built in blur, auto frame and eye focus modes for the webcam.

Webview2 with React to share code with the Web versions.

In a complicated Excel sheet I had the Conditional Formatting Rules Magaer crash due to an issue in EmbeddedWebView2.dll. Renaming it helped preventing the crash without obvious decline in functionality in Excel, until I tried to join a Teams meeting which obviously uses WebView2.

Even XBRL caved to JSON with XBRL-JSON. XBRL, of all standards.