Forge36

I love my e-bike. It sounds like you bought an electric dirt-bike. I'm not going to call you an idiot: the market is confusing and not well regulated, manufacturers are taking advantage of this. There have been some improvements with regulation to promote safety (both for you, as the unaware rider and for others). It's possible where you live it's not legal to take your new e-bike on the road without a motorcycle license OR on bike paths. Be safe, double check your local laws. If it's not legal: return it.

Wi-Fi was never unreliable. The devices that used WiFi were much worse to work with and setup.

Having switched to zigbee devices recently I regret trying to start with WiFi. While I'm not sure how it looks under the hood, adding devices with home assistant was miles easier.

Can you elaborate?

Location and time

That's known as Transmitter Hunting (aka Fox Hunting).

It's a fundamental flaw of most wireless communication. Are you referring to something more specific?

There something amazing about that. Thanks for pointing it out!

On a recent project we joked "developers can't read". Occasionally we'd ask for help and be pointed to the docs "I can't read".

I suspect there's two big parts to this:

1. Users expect batteries included and that everything "just works" the first time.

2. The language you used differs match your audience. E.g they search "gray" and find no results, however you've spelt it "grey"

Profiling[1] your own repository and tweaking as necessary (possibly disabling auto-status refresh) will likely yield significant performance improvements.

I use magit with a very large repository (100k files, millions of commits) it's still not lightning fast like it is with smaller repositories, I'm still finding it an improvement over the CLI.

My config notes this saves me ~13 seconds in git-status

  (remove-hook 'magit-status-headers-hook 'magit-insert-tags-header)

[1]: https://docs.magit.vc/devel/magit/Performance.html

Bikes typically have a weight rating. Mine has 150lbs of cargo capacity (is a cargo bike).

Context switching like that is exhausting

For me: Will this task take 30 seconds or 3 minutes. With good planning I've been able to step away and come back. Sometimes it decides to prompt me within 5 seconds for permissions. Sometimes it runs for 15 minutes.

The output is still small and I can review it. I can switch tasks, however if it's my primary effort for the day I don't like stepping away for an hour to do something else.

I'm still writing code. I'm doing it to solve a problem, there's more to writing code than than typing. Recently AI massively simplified "getting started", and all of the tips here are applicable to working well on a team.

My recent experience: I'm porting an app to Mac. It's been in my backlog for ~2 years. With Claude I had a functional prototype in under a day getting the major behavior implemented. I spent the next two weeks refactoring the original app to share as much logic as possible. The first two days was lots of fun. The refactoring was also something I wanted to flush out unit tests, still enjoyable.

The worst part was debugging really bugs introduced to my code from 5 years ago. My functions had naming issues describing the behavior wrong, confusing Claude, that I needed to re-understand to add new features.

Parts of coding are frustrating. Using AI is frustrating for different reasons.

The most frustrating part was rebasing with git to create a sensible history (which I've had to do without AI in the past), reviewing the sheer volume of changes (14k lines) and then deciding "do I want my name on this" which involved cleaning up all the linter warnings I'd self imposed on myself.

Menards has done this since at least 2012

Years ago (2015?) I bought a DS416j. Paid maybe $400 at the time. It was the simplest plug-n-play NAS could I could find. I didn't even know it had a GUI! (I was being very lazy with my searching). Two 2tb HDDs. Eventually I started running home assistant (no docker). 2023 I was getting annoyed with performance. Two 2tb SSDs added (4 slots) which helped. 2025 I've moved home assistant to a dedicated $70 thin client.

As a dedicated personal backup for my family it's been perfect. The latest vendor lock in has me reconsidering how I'll upgrade when the time comes. Until this post I was considering a $1000 unit and transferring my SSD drives before buying more storage.

Sounds like it's time to build a proper storage+computer rack.

I moved out of keep. It's an HTML backend.

The history of ethanol in gas is quiet old.

Per https://www.eia.gov/kids/history-of-energy/timelines/ethanol...

It looks like it usage of ethanol as an additive started increasing around the same time lead was being phased out.

Isn't a large part of ethanol it's use as a fuel additive that it boosts octane and is relatively cheap? Compared to leaded gasoline it seems very "green".

A statement that writes itself? Neat!

The default rarely change avoid breaking changes.

It makes for a good test

I've found targeted agents work best. Of course this also means I now have ~8 different tools I can choose to search with.

They aren't generic and the output needs to be monitored.

I couldn't find the tester. Thankfully the client i was tested... And it behaves poorly. Thankfully emacs has a client I can switch to!

I wonder if that's lost with this paper's method of typing.

I learned on a Mac. The course material was copying from a book without looking at the screen, then correcting the errors. This second pass to correct mistakes is different, and it's not clear that you can be removed when taking notes without affecting learning.

They give a recommendation for education, but didn't measure retention.

I agree with their premise provided students don't know how to touch type (given the anecdotal amount of visual typing I've seen a reason concern for students) it's also interesting that it's cursive, which is being taught less.

I think it gets worse > To prevent artifacts produced by head and eye movements caused by shifting gaze between the screen and the keyboard, typed words did not appear on the screen while the participant was typewriting.

So no only were typists not touch typing, they couldn't see what they were typing.

Too many variables are changed between the two tests.

A) writers can see whole words vs no visual feedback B) writers use their natural writing technique (2 or 3 fingers?) vs 1.

It'd be interesting to test typing with a single hand (using words that can be entirely touch typed with one hand)

I'm surprised the choose not to gather that data.

https://repository.lsu.edu/cgi/viewcontent.cgi?article=2306&...

Is interesting in that it shows the opposite effect: improved recall. It also measured velocity/speed of response. Which shows writing was faster.

Another idea for a test: writing without ink.

Magic smoke is released. Do try to avoid doing this with headphones in ears.

Commits aren't always perfect.

Sometimes I'll make the unit test first, which fails CI and the next set of commits implements the behavior.

I grew up with my father's 84' Ranger. 2 seater. Everything else looked big by comparison. And many had a smaller bed!

Legally it depends (I am not a lawyer)

The DOJ is likely to pursue this as collusion[1] which had legal precedent.

In this case: both parties intentional gathered to review prices together (through use of a third party).

Some bits from the article RealPage allowed landlords to "share confidential data" and "to suggest rents and term".

I don't know to what degree suggestions were made. If they were consistently suggesting increases: It's may be less similar to "looking at your competitor" and more "competitors agreeing to continuously notify each other of a price increase"

If it's also sharing occupancy statistics notification of an inelastic market could also be a contributing factor.

[1]: https://www.justice.gov/opa/pr/assistant-attorney-general-jo...