Deploying the code, see the bugs, fix the bugs, deploy more code. Best way to fix what the AI doesn’t see is a live environment

Always funny to me how much “time” in queries makes things more complicated, snapshot in time? Sure my time or your time, or are we using Zulu time? Or you wanted the time it was generated not the window it was forecasted?

Never fear an iframe

Optimize for the “sql query to production real time dashboard” metric

Some days you’re a Pokémon master catching every bug. Other days they’re Ghost-types and you’re haunted by the ones you can’t find.

You’ve heard of bike shedding but have you hear of AI driven bike shedding? We’re in the future

Been telling the AI “imagine a shrewd professor willing to merge changes but only if they are backed up by truth and reason, the professor likes ASCII art and has only limited time to read so description must be focused” when I need a PR description, great success.

It’s probably shouldn’t be surprising, but I do find it funny within 48hrs of moltbook being live the AI bots are already super pro privacy, E2E encryption, and ideas like “there’s the town square, now we need a backroom”

You can tell when Claude code became good, I should probably take a break soon

damn of course GitHub actions have an outage

wooo baby, I swear, changing the internals of an eventing system is peak stress

A thing few realize until they get into the weeds with real data, you want to index it but *just enough* letting the application/real usage guide you where an index should be applied or removed. Too many indexes can really mess up your write times. This is where taste comes in

I always get a little excited where I find a real live code spot that an atomic value actually is needed, it’s like a little reward for running a prod workload with real users