All issues

Cold-start latency spike on the canary fleet after deploy#3006

New issue
Open
NONoor Hassancommented 6d ago

Is your feature request related to a problem?

Cold-start latency spike on the canary fleet after deploy — today the workflow on praxis requires an out-of-band step that breaks the editor flow and makes the API harder to type.

Describe the solution you'd like

Migrate the affected surface to follow the same contract as the rest of the toolchain, so composed flows stay consistent and predictable.

Describe alternatives you've considered

  • Keep the current implementation and document the seam.
  • Ship a codemod that rewrites call sites at install time.

However, both would still leave the surface on a different contract from the rest of the stack, and would only delay the cleanup.

Labels: refactor

NOnoor-hassanself-assigned this6d ago
NOnoor-hassanadded the labelrefactor6d ago
NOnoor-hassanadded this to the milestoneQ4 reliability6d ago
NONoor Hassancommented 5d ago

Re-reading the spec, I think the right contract is the one in the bug report, not the one we're shipping. Going to open a small PR.

NOnoor-hassanadded a commit that references this issuefix(praxis): wire the new contract for #30065a6c8f45d ago
NOnoor-hassanmentioned this in#30545d ago
NONoor Hassancommented 5d ago

I've seen this too. I'll fix.

NONoor Hassancommented 5d ago

Closing in favor of #3107 — same root cause, same approach, and that branch is further along.

NONoor Hassancommented 4d ago

Can confirm the repro on a clean checkout — the failing case is reliable on macOS 15.2 with Node 22.

Add a comment

M↓Markdown is supported