User Manual

28. Logs, support copying and performance interpretation

Open Logs → Activity & diagnostics. Summary counts distinguish errors, warnings and retained activity. Use All, Errors, Warnings, Inspection, Marking and Publication to narrow the view, then Search activity to filter module/message text. Filtering does not erase history or rerun an operation. An empty filtered view is not proof that no earlier issue exists.

Support action Effect and limits
Run self-check Checks panel DOM, host bridge and active-document readiness without a full content inspection. It does not certify all detector behavior.
Copy last issue Copies the most recent retained warning/error as plain text, including the original technical message. Review it before sharing.
Copy document report Copies current single-document diagnostic JSON with its minimized content contract.
Copy last document report Can expose a matching retained pre-Apply report when no current result exists; it is not a new post-change inspection.
Clear history → Confirm clear Clears retained diagnostic messages and timings, not document text, profile libraries or publication data. Capture needed evidence first.

Compare the right timings

Last inspection includes extraction, analysis/result preparation and UI overhead. The status-bar analysis time need not equal its full total. Last optimization can show candidate pipeline counts, cache status, exact evaluation and rejection/near-miss evidence. A near miss is diagnostic, not an approved candidate.

Marking distinguishes freshness checks, preparation, condition setup, range resolution, native writes, verification/readback and canvas refresh. Verified marks reused / zero writes means the existing owned marks were validated and reused; it does not mean no validation happened. A host skip or unavailable anchor is different from a successful write.

For a fair performance comparison, retain the same build, document copy, scope, modules, profile, cold/warm state and whether you are measuring inspection or marking. A new cache state or a different sample is not a speed regression or improvement on its own. Sharing a plain log and sharing minimized diagnostic JSON are different disclosures; remove client-sensitive details from voluntary attachments.