GPT prompt:

"Based on the history of our conversations, suggest a few types of music that you think I'd love"

Hit reply if it delivered!

ChatGPT: Solves PHD-level problems

Also ChatGPT: You should totally mount a towel holder right next to your shower head.

Listing Bot - #1 for directory submission sites! Over 10,000 directories in db and growing in ListingBott! List SaaS, product, or website on 100+ directories in one click, saving you days of work to focus on more creative tasks. Listing Bot.

You might want to check out listingbott.com by @johnrush.me

Looks very neat!

I’m available for hire.

I’m an experienced Ruby, TypeScript and CSS engineer looking for a new full-time position.

I can help you with performance, concurrency, testing, architecture and API design problems on tooling or product teams.

I’d really appreciate any personal introductions. DMs open. 🙏

Interesting, why is that the case?

Coming from Ruby, I find benchmarking in compiled languages a lot harder.

In Ruby, what I write gets executed. In a compiled language, the compiler can compute what you benchmark in compile-time if you're not careful.

Always looks at the disassembly 😅

I really enjoy writing Rust in copilot. Not because it produces valid code (it does not), but usually it gives me clues on how to transform the current object into something a given function accepts that would've taken me a few minutes to get right without it.

Lets try an experiment. Should I post my fitness stuff here or keep bsky tech-related? Here's a random set 8 dips with +45kg. Ever since I started training, I dont feel any back pain from prolonged sitting 💪

#calisthenics #skygym

Most of the time I'd do smth like `flex-col md/lg:flex-row` and perhaps also center on xs screens. Things usually look nicer centered on small screens.

I have to go to sleep lol

Also, around 10ms is spent rendering SVGs 🙃 I can LRU cache these calls.

I'm not gonna do scientific benchmarks for now, but when profiling locally in production mode, it seems that it takes 4-5ms to render one element (MBP M1), and half of it is rendering the dropdown.

And the AR records are all loaded on the first element, so there's a "view penalty" from it as well?

No sleeping 😅 The Phlex part is embedded in a Rails view (the whole collection is a single Phlex component though).

It dances with Rails a bit. The dropdown is... a Rails partial that renders a ViewComponent, because I haven't yet migrated all to Phlex. I can profile and see what the bottleneck is.

The view that takes ~100ms to render is actually Phlex.

It renders a list of entities and has quite a bit of markup because each elem has a dropdown menu.

Here's what it looks like. The collection is uncached.

So I rebooted Intel and it works much better now. Still slower than AMD but 1.5x, not 4x 😅

Yeah it's a shared one

OK, I'm onto something. Changed my CPU from Intel to AMD and it starts looking reasonable now.

I'm gonna switch back to Intel once more - perhaps either loud neighbour or smth wrong with physical server?

Or is AMD really THAT MUCH FASTER?!

really nothing interesting here.

Works faster than production locally.

Hetzner is 2x slow as Render even though it has 4vCPU, while Render has 0.5 CPU for web and 0.5 cpu for postgres 😬

Gonna look further...

Good question! Gonna check it out in an hour

Its probably either a loud neighbour on the cpu or I really f'ed something up when upgrading Rails from 7.2 to 8.0 🤔🤔🤔

The DB is on the same Hetzner machine as my app

Same view on render.com

Soo I'm migrating my Rails app to Hetzner with Kamal. Running Rails & postgres on the same machine.

Even though the machine is much beefier than render.com, the views are rendering super slow. What am I doing wrong? 😬😬

Holy shit, that sounds like my life too, but I'm a bit less extreme. I force myself to finish (but usually I'm late for activities planned post-work), and I sleep a lot :D

But I really remember my anxiety kicking in in the office when ppl wanted to chat in the kitchen and my head was all abt code

Great stuff!

GitHub - zdennis/activerecord-import: A library for bulk insertion of data into your database using ActiveRecord. A library for bulk insertion of data into your database using ActiveRecord. - zdennis/activerecord-import

I've always used github.com/zdennis/acti...

would be smth like:

```ruby
Item.import([{id: 200, other_id: 100}, { id: 400, other_id: 200]] , on_duplicate_key_update: {conflict_target: [:id], columns: [:other_id]})
```

I would even prefer a single input with a locale-specific mask (eg. 11-08-1980)