Incremental addition, or additive accumulation is how I think of it.

1848 to 1900 is 52
1900 to 2000 is 152
2000 to 2026 is 152 + 20 (172) + 6 for 178

The steps can be smaller or bigger depending on how instant the mental math is.

Or should they? Easy skip if you’re sending resumes only to viable opportunities.

Or to target them to get an offer and decline with feedback that they sound like buffoons when they do such things. If you’re petty like me.

I’d at least wishlist Danzig, Danzig Revolution.

I've been updating the same gems. It's been surprisingly pain free for a good while now.

The worst has been Pagy since it went to v43, but just because all the methods changed.

I have anecdotal evidence, that's why I was sharing. In our massive, legacy test suite, rspec is slower than minitest. Our factories are ok, but maybe not optimized. Switching those to minitest with AR calls sometimes gets 4x faster, sometimes 0.

Rspec is slow in our context, not all contexts.

I only wanted to point out that your experience doesn’t warrant a blanket statement (IMO) that minitest can’t be faster than rspec universally. I apologize if it came across as an attack or attempting to change your mind or be combative.

It’s never been about which fw is faster. It’s about which is easier to achieve suitable performance. So it’s contextual, but that doesn’t mean it’s irrelevant.

And I know you’re not saying it is. But I think your spicy take is rather local, Mr. HasGoodFactoriesAndATopHat.

That’s probably true. But we’re smart people doing our best, so I don’t think it’s incompetence. Maybe it’s that you need to know more to write good factories.

So then using a framework that doesn’t put that burden on us is significant.

We’ve seen gains up to 4x changing a spec with factories to a test with fixtures and AR calls (1:1 parity, no other optimization).

Now, that might be because our factories suck, so the root cause is human. But it does matter if you have a mature suite and specs take more than a few seconds each.

Hopefully the value offsets the expense into the negative.

If I ever had to leave vim/Neovim, Zed would be my main.

But when all the pros and cons are weighed, (neo)vim still wins by a fair margin.

I agree. Which is weird because your dogs win. That’s just the way dog logic works, no need to question.

A Great Pyrenees runs in the snow with her tongue flopping across her nose.

Also, there’s a third contender named Juniper:

The correct answer is: “yes”

Sniping on that map that was two towers separated by a bridge was my jam. Headshots for days, but it was a cheap tactic.

Never could use the rocket launcher like some of those guys.

Hell yeah. We did Unreal Tournament, and Conker’s on N64 had a multiplayer that we played until controllers broke.

Screen peeking was part of the strategy, eventually.

I would hope most authors put the most effort in the first few chapters. They've existed the longest, had the most editing and there's little upstream to prohibit changes. Unless the author has a track record of sustaining, I'd abandon anything that doesn't make escapism enjoyable. It's medicine!

I play every RPG up to the final boss for the same reason. But then 30 years later I finally beat it with only the single weakest character as a challenge run. Your move, game devs.

This is the way.

Same to you!

Ha! The math on that one has some decimal places. I’m on my third rewrite of a side project and I’m achieving the same results with roughly 75% less code, and clearer architecture. There aren’t many feelings quite like it.

I’m rarely certain. I’ll write a tidy abstraction for something and never leverage it. Or I’ll keep it local and need to extend it a day later.

It’s probably better to err on the side of local, but that’s how you get huge methods and long files.

I’d like to think I know, but my success rate is low

Shaka, when the walls fell.

If Tim Pope hasn’t already written a fix for it, we are all of us lost at sea.

Are you just trying to use yegappan/lsp?

I know of one tutorial that shows a working setup, but I switched to neovim basically purely because the LSP stuff was a little easier.

If you want it for Ruby work, I think it only supports solargraph.

I hope you still filed the claim. It's death by a million cuts, but each cut gets us closer to doing something these companies will actually feel. Or so I hope since one of those buffoons said the industry would collapse if they had to pay everyone they stole from.

27 x 10 = 270
27 + 27 = 40 + 14 = 54
270 - 50 = 220
220 - 4 = 216

Slow, but I tend to favor fewer chances for mistakes.

Doubling 4 times would also be an option, and probably selected based on mood. I probably would go that route if the non-8 number was a factor of 5.

I had no idea about the whole SmallTalk OS/environment. That seems both really cool and slightly insane.

noelrappin.com/blog/2025/11...

Smalltalk is often referenced, but rarely explained. I never had a chance to work with it in its heyday, so stuff like this is really cool to read.

@pascallaliberte.me - it's like @noelrappin.com sensed our curiosity emanating from Hotwire Office Hours!

10% means just move the decimal left. So 5 times 3 is easy mental math for 15.