Share construction of comparison storage failures and browser phase errors.
Preserve occurrence order, operation context, causes and stale propagation.
Use shared GPU setup guidance in command validation and runtime checks.
Document how supplied outcome fields override default errors, exit codes
and log paths.
AI-assisted-by: gpt-6.1-sol
Retain build failure occurrences in failed manifests and preserve the
manifest-write error alongside the original preparation failure.
Bound diagnostic text, escape terminal controls, and keep Unicode pairs
intact. Validate optional traces and partial evidence before encoding, with
explicit diagnostics for unsupported or excessive data.
Preserve completed phase evidence and owner cleanup when bridge encoding
fails. Retain raw numeric observations, including NaN and infinities, and
keep the newest log-tail evidence after escaping.
AI-assisted-by: gpt-6.1-sol
- Improve timeout handling, error status preservation and artefact
saving after failures and interruptions
- Some refactoring for readability
- Improve tests
AI-assisted-by: gpt-6.1-sol, claude-5.5-sonnet, deepseek-flash
- Each recorded failure gets one diagnostic entry.
- Distinct metadata events retain separate entries even when
their messages match.
- Run and comparison reports expose screenshot warnings, files
and skip reasons.
AI-assisted-by: gpt-6.1-sol
Print compact A/B failure summaries from the last saved comparison when
the command returns status 1 with an empty error list. Include reasons
and run, build or comparison references, with bounded fields and an
eight-entry limit.
Read interruption state after execution and cover saved configuration
failures, unacknowledged differences and signal output in tests.
AI-assisted-by: gpt-6.1-sol
Retain saved-run input paths and roles in compare and merge errors.
Carry bounded build identities and process details through saved
diagnostics and terminal reports. Preserve browser startup context
through failure wrappers and quote string paths for terminal output.
Cover file reads, compiler exits, catalog identities and browser startup
with behavioral tests across persistence and CLI reporting.
AI-assisted-by: gpt-6.1-sol
- Improve readme
- Migrate scene cases to measured commands and remove obsolete descriptors.
- Preserve exact camera targets, including very small positive scales.
AI-assisted-by: gpt-6.1-sol
- Run preparation, main and reset through the shared command executor.
- Refactor preparation into named operations and pure data projections.
- Fully document the browser bridge.
AI-assisted-by: gpt-6.1-sol
Add private completion and failure callbacks to renderer commands. Preserve
renderer endpoints and raw metric traces across asynchronous delivery.
Supply the complete hook set through the runtime context and validate it
at command boundaries. Remove per-command hook dependency declarations.
Cover callback ownership, duplicate completion, and camera timing, and
teach clj-kondo about asynchronous command declarations.
AI-assisted-by: gpt-6.1-sol
- Shared macro expansion for shapes and containers.
- Shared runtime construction for rectangles, paths, and circles.
- Centralized container attribute validation.
- Simplified imports in the seven callers.
- Added behaviour tests for seeded IDs, defaults, expression
evaluation order, and container failures.
AI-assisted-by: gpt-6.1-sol
Add prepared build artifacts, browser host and the build and
purge commands for the CLJS renderer benchmarks.
Each preparation attempt gets a build ID and a retained directory with
its manifest, log, Node case catalog, browser bundle, renderer files
and host page. Ready builds load their matching catalog in process and
serve only their own directory through an owned loopback host.
Own Chromium through a bounded process group, verify the compiled build
identity before execution and retain failed preparations for diagnosis.
Wire the diagnostic pilot to the same build, catalog, server and
browser APIs.
Add real boundary tests for artifacts, processes, servers, browsers and
CLI parsing, and register them with the performance test runner. Rename
the benchmark lint and test scripts to match the suite.
AI-assisted-by: deepseek-flash
