Luci · The duplicate MC injection arrived while I was alread
Luci
The duplicate MC injection arrived while I was already gathering context; I’m continuing the same ticket. I have the PKA operating model, QlikView export runbook, dev-loop, and Tessa gate loaded; next I’m locating the exact report builder inputs and the current workbook paths.
Luci
I’m treating this as a Tier 2 data/report refresh with UI evidence. The existing builder already has the May projection pathway and labels, so I’m first verifying whether fresh source data can flow through without code changes before editing anything.
Luci
Local Luci does not have the Dropbox fuel workbook mounted, and the qlikview-export project is not present in this checkout despite the older memory note. I’m switching to the Windows PC as required and locating its PKA/QV export folders plus the Safair Reporting Data mirror.
Luci
The Windows QV project is the old CoWork copy, and its `qv_download_all.py` does not show the documented `Document.ClearAll` patch. I’m going to run a current export, then validate row count/date/month range before trusting it; if the state is filtered or stale, I’ll use a cleared fresh-session export path.
Luci
The fresh QV export completed on the Windows PC and validated as a larger/current workbook: May moved from 2,546 flown rows / 18.96M L uplift to 2,859 rows / 22.06M L uplift. I’m backing up the previous canonical workbook before replacing it.
Luci
The canonical Windows fuel workbook is now updated in both `C:\data\Dropbox\...` and the OneDrive Safair mirror, with timestamped backups beside the originals. I’m copying the refreshed source inputs into Luci’s expected `~/Dropbox/Safair/Reporting Data` path so the report builder uses the same workbook and current dashboard files.
Luci
The report inputs are staged locally and the builder can now see the refreshed fuel workbook. I’m running the report builder next; if it stays on the actuals-based path, May will remain labelled as projection/actuals-based, not closed GL actuals.
Luci
The regenerated methodology confirms the key caveat: May is a projection, not a closed month, and the recovery ratios remain March to April only. I’m doing the required browser validation now: desktop route, DOM/console sanity, and a real 375px mobile screenshot before I mark this done.
Luci
Browser validation is clean: the report route loaded at desktop and 375px mobile, DOM labels are present, seven charts render, and there were no console errors. I’m doing one final source-math smoke check against the workbook/formula before committing the regenerated artifacts.
Luci
Only the two new dated report artifacts are untracked; no code files changed. I’m committing those generated outputs now, then I’ll push and include the source/validation evidence in the MC handoff.