Build the a6 binary from main, point it at a real APISIX deployment (proposal: a local docker-compose with APISIX 3.15 + etcd, plus at least one run each against 3.2 LTS and 3.9 for the matrix), and run the phased CLI walkthrough by hand. For each of 14 resources: create / get (all output formats) / list / update / delete / verify cleanup. Plus declarative config dump / diff / sync / validate.
This walkthrough is where real-traffic verification lives (route forwarding, multi-node upstreams, key-auth flow, etc.) per #14's earlier semantic cases. Capture every observation in docs/ga-test-report.md with pass / fail / bug columns. Each bug found gets a failing test added first, then a fix (a7 protocol). Re-run after fixes as Run 2 regression.
Primary source of bug-discovery for GA. Feeds the skip-elimination, UX-sweep, and skills-validation sub-issues.
Part of #33
Build the
a6binary frommain, point it at a real APISIX deployment (proposal: a local docker-compose with APISIX 3.15 + etcd, plus at least one run each against 3.2 LTS and 3.9 for the matrix), and run the phased CLI walkthrough by hand. For each of 14 resources: create / get (all output formats) / list / update / delete / verify cleanup. Plus declarativeconfig dump / diff / sync / validate.This walkthrough is where real-traffic verification lives (route forwarding, multi-node upstreams, key-auth flow, etc.) per #14's earlier semantic cases. Capture every observation in
docs/ga-test-report.mdwith pass / fail / bug columns. Each bug found gets a failing test added first, then a fix (a7 protocol). Re-run after fixes as Run 2 regression.Primary source of bug-discovery for GA. Feeds the skip-elimination, UX-sweep, and skills-validation sub-issues.
Part of #33