sync: auto-sync from AD2 at 2026-06-17 13:35:55

Author: Mike Swanson
Machine: AD2
Timestamp: 2026-06-17 13:35:55

Prompt617.txt

@@ -0,0 +1,65 @@
+  TASK: Diagnose + fully document the Dataforth test-datasheet bug. You are running ON AD2
+  (192.168.0.6), the host of the testdatadb generator + PostgreSQL. Do NOT fix anything yet <20>
+  trace, document, and pinpoint root cause.
+
+  === THE PROBLEM (reported today, 2026-06-17) ===
+  John Lehman (jlehman@dataforth.com) to Mike: "We are experiencing problems with test
+  datasheets. Column headers are wrong. Some Final Test lines are missing."
+
+  From the thread (apps engineer Peter Iliya):
+  - For the 8B35: the STAGED datasheet "H9553-13" (T: drive, TS-4L/STAGE) has the CORRECT
+    input column label "Temp" <20> but the WEBSITE version "179553-13" shows the WRONG label
+    "resistance". DSCA38 is flagged with the same issue.
+  - Peter CANNOT find 179553-13 as a file on the X: drive or in the DFWDS report <20> because the
+    website version is generated from the DB, not stored as a file.
+  - Core question (Peter): "Is there a reason we list the input for our RTD modules as
+    resistance, but show the temperature instead?"
+  - Trigger: customer Wellbore Integrity (Joseph Swinehart) raised a cal-cert AUDIT discrepancy
+    on 8B35 4-wire RTD certs.
+
+  === THE KEY POINT (Mike) ===
+  The rendered test sheet EXISTS AS A FILE BEFORE it is ingested into the database. That
+  original/staged file is the ground truth (correct Temp label). Our DB-based regeneration is
+  what introduces the error. FOLLOW THE TRAIL UPSTREAM and FIND WHERE THE ORIGINAL FILE IS.
+
+  === ARCHITECTURE (per TEST-DATASHEET-PROCESS.md) ===
+  Test stations (TS-01..27, DOS) write .dat logs
+    -> AD1 HISTLOGS: \\ad1\... C:\Shares\test\Ate\HISTLOGS\{log_type}\{model}.DAT (+ per-station \TS-XX\LOGS\)
+    -> Legacy DFWDS (VB6) historically rendered "For_Web" .TXT datasheets (the ORIGINAL rendered files)
+    -> testdatadb (THIS host AD2): Node.js + Express + PostgreSQL 18, service on :3000, dashboard
+       http://localhost:3000/, WinSW wrapper, service account INTRANET\svc_testdatadb
+    -> Hoffman Product API (/api/v1/TestReportDataFiles/bulk, OAuth2) -> public website.
+
+  === WHAT WE ALREADY FOUND IN THE CODE (verify the DEPLOYED copy matches) ===
+  Generator template: datasheet-exact.js (find the deployed copy on AD2; repo copy is
+  projects/dataforth-dos/datasheet-pipeline/implementation/templates/datasheet-exact.js).
+  - getSensorNum(): RTD -> 7. Input header logic: sensorNum 3-6 (thermocouples) -> "  Temp. (C)";
+    sensorNum 7 (RTD) -> " Rin (ohms)"; so RTD is rendered as RESISTANCE. The accuracy VALUE
+    formatting is ALSO sensorNum-driven (temperature format vs resistance format) - so values may
+    be wrong too, not just the label.
+  - Final Test loop: `for (i < dataLines.length && i < parsed.statusEntries.length)` with
+    `if (status.length <= 4) continue` -> rows DROP when statusEntries are misaligned/short.
+
+  === YOUR DELIVERABLES ===
+  1. FULLY DOCUMENT THE END-TO-END PROCESS (write it to a markdown file), especially the UPSTREAM
+     half: exactly what the test station emits, what the ORIGINAL rendered datasheet file is,
+     WHERE it physically lives (find TS-4L/STAGE, the staged H9553-13, and/or the DFWDS For_Web
+     .TXT), and exactly how testdatadb ingests it (parser, which fields, the .DAT format).
+  2. LOCATE THE ORIGINAL FILE for the 8B35 example (H9553-13 / its For_Web .TXT / the .dat it came
+     from) and for a DSCA38 example.
+  3. DIFF original (correct) vs DB-generated (wrong) for 8B35 RTD and DSCA38: the exact column
+     header difference, any VALUE differences, and which Final Test lines are missing and WHY.
+  4. ROOT-CAUSE LOCALIZATION: is the defect in (a) ingestion/.DAT parsing, (b) the DB data itself,
+     or (c) datasheet-exact.js rendering? Confirm against the original file, not assumptions.
+  5. State the CORRECT RTD output (label = Temp (C); values = temperature, derived how?) per the
+     original, and the precise code change(s) that would fix both the header/values and the
+     dropped Final Test lines. Note any modules where resistance IS the correct label (don't
+     over-correct).
+
+  Resources you have locally: PostgreSQL on AD2 (query testdatadb directly), the deployed Node
+  service + its code, SMB to AD1 (\\ad1\...) and D2TESTNAS (\\192.168.0.9\test, /data/test has
+  DFWDS, Ate, 8BDATA, DSCDATA, etc.). The ClaudeTools repo has the pipeline source under
+  projects/dataforth-dos/.
+
+  Output: the process doc (file) + a diagnosis summary + the proposed fix. Diagnose only <20> no
+  changes to the generator or DB until reviewed.