Christoph Lupprich

❤️

Exhibit #52636153 of cold emailing gone wrong 😂

Yeah, it‘s a disgrace. They should go back and focus on their core product.

Don't throw the specs out with the factories FixtureBot gives you the speed of fixtures with the syntax of factories

I got tired of seeing people complain “RSpec and factories takes 14 min to run, minitest and takes 5s” so I’m starting to chip away at it.

First order of business: created FixtureBot after figuring out I use FactoryBot not because of factories, but because I like the syntax.

Might try sus next.

This is awesome - thanks!

I mean, they didn’t deliver any true innovations for quite a while now (not saying that maintaining things isn‘t already hard). So sounds like it‘s more like a realization of their status over the last couple of years?

Oh wow, what a bummer! I‘m running two hobby projects on @railway.com and I’m very satisfied.

GitHub - kaspth/oaken: Fast tests, clear datasets. Ditch nightmare fixtures and slow confusing factories. Fast tests, clear datasets. Ditch nightmare fixtures and slow confusing factories. - kaspth/oaken

You might wanna have a look at github.com/kaspth/oaken

Been tinkering with a little app to get insights into the performance of your GitLab CI pipelines, mainly to spot degradations and verify that improvements are actually… improvements. Curious if anyone else would be interested in it?

A weekly time budgeting dashboard showing availability across multiple projects and days

So I built my own solution: Crow's Nest.

It's a dead-simple dashboard that connects to your Harvest account and shows:
- Your total weekly capacity.
- Hours you've committed.
- How many hours are actually left.

No more guessing.

GitHub - ash-project/usage_rules: A tool for synchronizing LLM rules files with your dependencies. A tool for synchronizing LLM rules files with your dependencies. - ash-project/usage_rules

Ah, found it: github.com/ash-project/...

Of course I can‘t find it right now, but I believe the Elixir folks made something that merges AGENTS.md files of dependencies so LLMs know what they can work with.

I have no idea what systems those airlines are running on, but they all feel horrible.

You mean something like Mailgun‘s templates? Ideally I‘d also like to replace SES with a service that operates within the EU. www.scaleway.com/en/transacti... sounds promising.

Took running a little easier over the last couple of months because it was quite cold, slippery, and honestly my mind just wasn‘t in it.

Did hard runs on Wednesday and today, and while I‘m pretty exhausted now 😅, it feels great.

Pre-Deploy Command for Multi-Service Rails apps on Railway Pre-Deploy Command for Multi-Service Rails apps on Railway - bin_release.rb

Ran into a race condition on @Railway.com when deploying Rails with web + Solid Queue worker. Unlike Heroku's single release phase, Railway runs pre-deploy commands per service.

Solution: wrapped the migrate in a wrapper script to
erialize migrations across services: gist.github.com/clupprich/a...

😅

Worked flawlessly, thanks! Interestingly enough they also seem to be using Amazon SES (based on DNS records).

What‘s everyone using to send transactional emails these days? Still Mailgun? I heard that Postmark is back, too? Please don‘t say AWS - I just hate their UI.

I moved a #Rails app from Heroku to @Railway.com yesterday and just created a new one there. Very solid experience, guess I found the new home for my projects.

Solid choice on the Garmin - I love mine. No recommendation on the headphones unfortunately, I don‘t wear any (anymore) while running.

How a Postgres Sequence Replaced Redis Locks at 90M+ Units/Year When Redis locks started deadlocking at scale, I found a simpler solution: let Postgres handle it. Here's how a database sequence eliminated our distributed locking issues and delivered sub-millisecond performance.

First blog post in… a while (201?).

Replaced Redis locks with a Postgres sequence at 90M+ units/year. Sub-millisecond performance, zero deadlocks, simpler architecture.

Sometimes the best solution is using what you already have.

christoph.luppri.ch/posts/postg...

I‘m using inwx.com, believe they are German.

TIL SolidQueue lets you schedule a single line of Ruby inline in the config file - no separate job class needed. 15 years in and Rails keeps finding ways to delight.

Two small improvements that would make my GitLab workflow smoother:

- Preserve form state when changing the base branch while creating an MR
- Check if an MR is still mergeable when the base branch changes - not only when someone views it

Coming from Rails, the lack of a proper console is my #1 Python pain point. Django’s manage.py shell gets you halfway there, but what are FastAPI/Flask folks doing? One-off scripts? Custom REPLs? Suffering?

omg, the memories!

I've got a surprise for hotwire.club members coming up 🥁

... an MCP server containing the entire knowledge base of challenges and solution up to date!

There will be a free version, don't worry.

The full version will be limited to DC members OR available for purchase from the shop.

Haven't been running for a week (wisdom teeth extraction + a cold with fever), and oh boy do I miss it.

Continuous deployment becomes unremarkable once it becomes routine, which is exactly the point. Happy you're living the dream! 😅