That's a good tradeoff I suppose. I've been racking my brain trying to find a solution recently that solves both of these but haven't been able to.
What I had landed on was idempotency on a best effort basis and just made the event processing safely retryable without violating any system invariants.
Too much work. Don't try to act as the sender yourself. The outbox pattern leaves that as an exercise for the reader. How many receivers do you have? Do they come and go while you're trying to send messages? Do they need to find out about old messages that were send before they came online?
There is much better alternative than this motte-and-bailey argument of "outbox GUARANTEES blah for a distributed system - but only within a single node".
Just write down what happened in Kafka. N followers read from Kafka to find out what happened.
Kafka is actually distributed tech. You can lose nodes and keep operating.
No need to design for atomicity. You either wrote to Kafka or you didn't.
I explicitly make sure services I lead have Integration tests in CI pipeline to validate the "negative paths" against all APIs with missing, invalid, un-authorised identities, expired, un-authenticated tokens. Of course that still doesn't cover every surface, but even that gets sideways glances from some folks who think we should just test happy paths and why we're testing for access controls in Integration tests.
Realized this during a particularly stressful time in 2021 - back then, I used to spend hours walking just thinking through problems, all night long. I’ve since abandoned the all night long part, but have an almost daily ritual to walk around thinking about whatever problem - small or big - I’m working on at the moment.
I’ve also found that during these walks, the more I talk out loud to myself and move my hands as if I’m writing on a whiteboard, the faster I get to an answer.
Been buying mullvad for the last 4-5 years but oftentimes I can’t even browse the fucking New York Times website due to low bandwidth, let alone stream anything. At this point, I just keep adding time to my account just in case, without using it.
> jack up premiums by 20% unless you force employees to change their password once every 90 days"
Always made me judge my company's security teams as to why they enable this stupidity. Thankfully they got rid of this gradually, nearly 2 years ago now (90 days to 365 days to never). New passwords were just one key l/r/u/d on the keyboard.
Now I'm thinking maybe this is why the app for a govt savings scheme in my country won't allow password autofill at all. Imagine expecting a new password every 90 days and not allowing auto fill - that just makes passwords worse.
> AI to run in loop, fixing compile errors, fixing tests, do builds, run the app and do API calls...
Ah I really wanna trust AI won't "fix" the tests by commenting out the assert statements or changing the comparison inputs willy-nilly. I guess that's something terrible human engineers also do. I review changes to tests even more critically than the actual code.
Oh the sha-256 hashes are precisely what I used for a quick script I put together to parse through various backups of my work laptop in different places (tool changes and laziness). I had 10 different backups going back 4 years, and I wanted to make sure I - 1) preserved all unique files, 2) preserve the latest folder structure they showed up in.
For better or worse, I have a habit of clicking the touchpad or a few keys after I shut down my laptop. Just to make sure it's shut down properly. Back in Windows days with HDDs and hibernate, laptops sometimes took minutes to shut down completely, and I don't like closing the lid before shut down is complete.
Now, I end up restarting with that mere act, and have to long-press to shut down again because the shut down option won't show up on login screen.
What I had landed on was idempotency on a best effort basis and just made the event processing safely retryable without violating any system invariants.