Toby Wint_rmute

69 Followers
86 Following
106 Posts
Software developer since the 1990s, enjoys rock climbing, motorcycles, techno and prog, among other things.
I migrated from hachyderm.io - older posts are still there.

Went back to Araps for Easter!
Trad lead on six routes (up to grade 13, most two pitches), trad second on four routes (up to grade 19) including the epic Watchtower Crack multi.

I don't think I posted on here about it, but I was at Araps last weekend too. Leading on Panzer (gr12 multi) and some fun single pitch routes.

OK i can't focus on work and keep looking at this repo.

So after every "subagent" runs, claude code creates another "agent" to check on whether the first "agent" did the thing it was supposed to. I don't know about you but i smell a bit of a problem, if you can't trust whether one "agent" with a very big fancy model did something, how in the fuck are you supposed to trust another "agent" running on the smallest crappiest model?

That's not the funny part, that's obvious and fundamental to the entire show here. HOWEVER RECALL the above JSON Schema Verification thing that is unconditionally added onto the end of every round of LLM calls. the mechanism for adding that hook is... JUST FUCKING ASKING THE MODEL TO CALL THAT TOOL. second pic is registering a hook s.t. "after some stop state happens, if there isn't a message indicating that we have successfully called the JSON validation thing, prompt the model saying "you must call the json validation thing"

this shit sucks so bad they can't even CALL THEIR OWN CODE FROM INSIDE THEIR OWN CODE.

Look at the comment on pic 3 - "e.g. agent finished without calling structured output tool" - that's common enough that they have a whole goddamn error category for it, and the way it's handled is by just pretending the job was cancelled and nothing happened.

I'm struggling to fix an electronic circuitboard -- can anyone offer suggestions?
It's a circuitboard from a motorcycle instrument panel, manufactured in 2019. It has a layer on each side. I don't know if it has any internal layers.
There are two push switches, and the controller seems to think one of the switches is continually pressed.
Testing the switch with a multimeter, it looks fine. It's normally open, and when pressed, closes the circuit with very low resistance. This is identical to the other, working, button.
The circuitboard has some test points, including for GND and the switches.
When my multimeter is connected to GND and the test point for the bad switch, the results appear good.
ie. No connection normally, and then a very low resistance connection when pressed.

I'm trying to work out what else could be the problem. Maybe the connection to the microcontroller has shorted out against a neighbouring connection?
I can't visually see anything like that, but I don't have a circuit diagram and the circuitboard has a layer on top of it obscuring the tracks.

Maybe a track has somehow broken inside the circuitboard and shorted to ground? I can't see any visual damage, and I guess I wouldn't know how to fix this even if there was.

Are there other possibilities I'm missing?

I haven't been down to the dodgy end of the CBD in ages! Jumped off the tram this afternoon and immediately saw a homeless guy punching out a tourist who had taken a photo.
Stay classy Melbourne.
Might continue to stay in the inner north hey

PSA: The Amazon wishlist doxing threat is much greater and more immediate than folks might realize. Attack works like this:

Stalker who wants your address opens an Amazon seller account and lists themselves as a third party seller for any item on your public wishlist. Then, they order the item from themselves as a gift for you. Bam, they have your address.

In particular, attack does not depend on an existing third party seller having poor PII handling hygiene, like the articles have implied.

Is VicEmergency trying to supplant the BoM for weather?
I'm getting a red shelter-in-place warning because of an incoming... thunderstorm?
YOU MUST GO INSIDE TO BE SAFE!

I'm pretty sure I've survived rain before. Or how bad is this about to be??

i’m so old, i remember when computers got cheaper over time

What on earth is causing this error?
New install of Home Assistant, works normally in Firefox, but Chrome is rendering all text shifted by two characters, so "Welcome" becomes "Ucjamkcâ".

Even stranger is that viewing the source in Chrome shows the correct text, and copying and pasting it also results in correct text.

Any ideas?

Update: Changing the CSS font in the browser debugger to remove Roboto fixes the issue. So does uninstalling Roboto from the computer.
Still confused about why this only affected one browser and not the other, though!

Check out this guy!
Came by while I was climbing, didn't seem scared at all of the humans. Just scrambling around, poking his nose everywhere
Is Google Calendar's timezone support totally broken or am I using it wrong?
While I'm in the +10 timezone, I create an event and say that event will be at 14:00 in the +9.5 timezone.
I then travel to the +9.5 timezone.
My calendar updates and shows the event at 15:00 local time.
WTF?