It's fascinating working with Claude on novel side projects and at work with lots of strict guardrails, AI bot reviews and cookie cutter patterns. On one side project Claude is completely untrustworthy, makes tons of mistakes, constantly gets confused, at work it 100xs me
@dinasaur_404@Cloudflare make it cheaper to deploy all workers in cd all the time with some sort of checksum to skip deploys
fix the slow loading save rather than deploy button
programmatic rolling rollouts and rollbacks
@MikeNomitch_CF @yagiznizipli Thanks Mike. 10 max instances where most are in the EU but some aren't, meaning it's difficult from a gdpr perspective. I currently have a hack where I init the container, ping it, return the region envar and if it's not WEUR don't send traffic to it and create another....
@WalshyDev@CloudflareDev Greater timeouts on durable objects, uncapped fetch calls from workers, proper dashboarding on worker analytics, checksum on worker code to skip deploys, programmatic gradual rollouts with autorollbacks based on error metrics, durable object instances in dashboard
@dok2001@Cloudflare woud love
* cross worker rpc in local dev
* queues in local dev
* workflow vitest
* out-the-box local dev for frameworks like solid start
* notification iac
* iac all round, especially around kvs and d1
* remove limits on api calls from workers
* better otel experience with logs