This note captures the first wave of post-adoption DX feedback.
The important conclusion is that Observer does not currently need conceptual redesign.
It needs smoother first-path affordances.
- Sharpen simple vs full suite surface diagnostics.
- Add ctx-taking C helper macros so helper functions are not forced to depend on
observer_test_ctxin direct scope. - Make selector-driven provider suite examples more prominent than exact-name memcopy examples.
- Build a cross-provider conformance suite that every first-party provider library must pass.
- Lock the provider protocol and canonical artifact compatibility policy in writing.
- Improve workflow and action failure context with clearer producer, path, publication, and assertion diagnostics.
- Add a
doctoror lint command for common authoring mistakes and suspicious configurations. - Add a lint for exact-name items that only restate a matching selector-level
expect exit = 0suite. - Define and document a first-class golden maintenance workflow for regenerating and reviewing canonical artifacts.
- Define a first-class product certification contract for products that require multiple local suites to pass together.
- Add one standard bootstrap command or template flow for new provider projects.
- Add an
observer fmtcommand so large.obssuites and workflow scripts can be reformatted mechanically instead of being maintained by hand. - Reduce provider workflow ceremony in the common case without weakening the explicit artifact model.
- Keep validating released surfaces on Linux and WSL2 so first-path onboarding, scaffolds, and everyday repo workflows do not quietly stay Mac-shaped.
- Add a product certification CLI and canonical product report artifact that execute ordered stage runs from explicit per-stage working directories.
- Add conformance fixtures for heterogeneous product certification, including unit plus workflow certification in separate local Observer areas.
- Keep strengthening the most visible examples so the intended copy path stays selector-first and inventory-driven.
- The workflow/artifact model is holding up well.
- The provider boundary is holding up well.
- The current rough edges are mostly discoverability, parser guidance, and helper ergonomics.
- Linux and WSL2 now matter for the released path, not just Mac development.
- The JavaScript shell-proxy scaffold had a real portability bug from a hardcoded Mac-style
PATH; that was fixed by templating the current machinePATHinto generated configs. - The repo-owned CMake dogfood path currently shows Linux golden drift, but CMake remains unreleased and should be treated as pre-release parity work rather than as a blocker for the released Linux surface.