Changes

Offer/contract e-mails: button is 'Angebot öffnen' / 'Dokumente öffnen' (view only, non-binding) — accepting or signing is described as a separate, optional step on the page; default message texts, info boxes and modal label reworded accordingly

Mobile movement: opt-in billable ticket after completion — post-completion screen with selectable moved articles/qty, time or direct amount ([Zeitabrechnung]), typed or voice-transcribed note (new sync /api/mobile/transcribe-text), ticket for the partner linked to the source locator's org, audio attached to the ticket

Mobile app release channel moved into the logyou.de website repo: new scripts/stage-mobile-update.sh copies APK + version.json into ../logship/logyou-landing/public/logship (archives the replaced APK into history/, merges history.json, keeps last 10); release-app.sh hands off to it (MOBILE_PUBLISH=sftp keeps the SFTP uploader for Blut24, --push also deploys the site); release app 1.0.40 (code 41)

Mobile app release channel inside the ERP: APK + manifests live server-side in data/mobile-releases (never in the bundle or git), served via expiring HMAC-signed public download URLs (/api/public/mobile/<token>/…, Range support); app-version/app-versions read the local channel for authenticated checks and keep proxying www.logyou.de/logship for old builds (MOBILE_APP_LEGACY_FALLBACK); new Settings → Mobile App page with QR install code, download link and admin upload form; publish endpoint (admin verified via ad_role or service token) + scripts/publish-mobile-update.sh; release-app.sh publishes to erp,landing; deploy script preserves data/mobile-releases

show bom on mobile search

release app 1.0.41 (code 42): movement ticket + BOM in mobile search

release-app.sh: interactive release — asks for versionName (Enter = auto bump), release notes, publish target (default: ERP channel only; landing/sftp optional, website push asked separately) and a final confirmation with summary; --yes runs non-interactively with the defaults

Mobile article search: BOM / JTL-BOM tags on results, "Stücklisten (BOM) ausblenden" filter checkbox (default off, on all 7 pages incl. receive), article names clamped to 3 lines, viewport-lazy product thumbnails (MobileProductThumb + GET /api/mobile/product-thumb/{id}, per-product client + 10-min server cache) — search endpoints return isBOM/isJtlBom/imageUrl

Mobile app update check now authenticated: useAppUpdate uses relative /api/mobile/app-version(s) so the Capacitor fetch wrapper injects the Bearer token (absolute URLs bypassed it and only ever saw the legacy www.logyou.de manifest); the raw-fetch wrapper also handles absolute apiBase/api URLs; release app 1.0.44 (code 45)

release app 1.0.45 (code 46): authenticated update check

Track .claude/settings.json so cloud sessions default to bypass permissions

Product Located-at tab: quantity pills aligned with their column headers, Active column removed, Date Material Policy shown as dd.mm.yyyy

Product form: hide Summary Level checkbox (commented out, not needed for now; isSummary stays in the model)

Transactions page: product search on the Nuxt UI SelectBox wrapper (tedir-select dropped) with rich results (thumbnail, article no., SKU/MPN/category, BOM + org + inactive tags), selected-product card with product link, ?productId= deep link that preloads the product and its history; PDF report extracted into useTransactionReportPdf; product search returns org/category/active

Product Transactions tab: order + shipment links (server resolves M_InOutLine → M_InOut → C_Order with a batched lookup; sales vs purchase routes), dates as dd.mm.yyyy with muted booking time, coloured movement-type icons and signed quantity pills, Active column removed, Report button (PDF preview modal + download) and link to the prefilled transactions page

iDempiere fetch layer: no ofetch auto-retry, 90 s timeout, and no token-refresh retry after backend errors

fetchHelper and event.context.fetch now send retry: 0 (ofetch retried every GET once on 500/502/503/504) plus a timeout (IDEMPIERE_FETCH_TIMEOUT_MS, default 90 s) and record the last iDempiere error on event.context. refreshTokenHelper throws instead of refreshing when that error is not 401/403, so the standard catch → refresh → retry route pattern no longer re-runs a query that just timed out on a stalled backend (2026-09-10 retry storm: 6 → 79 calls/min of one heavy query, 178 of 200 Jetty threads blocked).

Products with stock: drop $expand=M_Storage (global-lock hazard in idempiere-rest) for a paged m_storage join

New server/utils/productStorageHelper.ts fetches products plainly and joins m_storage rows in Node (by-id chunks for a page, all rows with a 30 s single-flight cache for full lists), keeping the M_Storage array shape. A detail expand runs DefaultQueryConverter.convertStatement (synchronized singleton) once per parent record — 3.3k products = ~6.6k serialized queries under one lock that every other $filter request waits on. Switched: materials/products (also gains ?top=&skip= and a 60 s per client/role/org response cache with single-flight), ag-grid products, qty-on-hand overview/by-locator/pit, locator priority recommendations.

Synology helper: timeouts on every NAS call, shared session, re-login on session errors

Every Surveillance Station request carries an AbortSignal timeout (15 s API, 5 min download) so a hung NAS cannot pin a request. One login is reused for 30 min across requests instead of login+logout per call; error codes 105/106/107/119 trigger exactly one re-login + retry with the session updated in place. synologyLogout is a no-op unless force=true, so the route handlers are unchanged.

Recipes: iDempiere REST load rules (no child $expand over big lists, no blind retries, $top everywhere) and the Synology session/timeout conventions

Mobile Reconditioning page (/mobile/reconditioning): scan return label (RMA no. / customer-return doc no. / product code) → customer return with lines on non-shippable locators only, select articles + qty (capped by on-hand and already-reconditioned qty), target locator with org default preselected, books an M_Movement linked via M_Movement.M_Customer_Return_ID (fail-soft if column missing), then optional billable 'Reconditioning' ticket (time/amount, articles as message, no note/voice); endpoints /api/mobile/reconditioning/{lookup,target-locators}; dashboard tile + en/de/es/ne translations

release app 1.0.46 (code 47): mobile Reconditioning page

BI Product Performance (/bi/product-performance, admins only): per-merchant product sales analytics — rolling window (3d…12m) with previous-period deltas or 12/6/24-month matrix (heat cells, sparklines, 3M-vs-3M trend); sold/returned/return-rate/net revenue/on-hand/coverage per product; scope = top N sellers or manual pick list (persisted per org, pin from table); KPI cards, daily and monthly charts, category share, worst return rates, CSV export; direct-Postgres route server/api/bi/product-performance.get.ts; sidebar + top-menu entries, en/de/es translations

BI Product Performance: hide charts/table until a report is loaded (no stale chart data when switching org or with an empty manual selection), subtle row action buttons

Mobile Reconditioning: lookup by RMA number only (no return-doc-no / product-code fallback), return label barcode now encodes the RMA number (was product SKU, so scanning never hit the RMA path), keep-alive focus lifecycle for the hidden scanner input, en/de/es hint texts

Mobile Reconditioning: A4 movement report (Umlagerungsbericht) printed after booking — server-rendered pdfmake sheet grouped by target locator (article, qty, from → to, checkboxes, disposed lines, signature line) sent via lp to the remembered paper printer (CUST_Printer 'pap', default-on toggle + printer select persisted per device); per-line 'Entsorgen' option books non-reconditionable articles OUT via physical inventory (MMIPI, qtyCount = on-hand − disposed, description documents the disposal) instead of moving them, target locator only required when something is moved; ticket + report cover both; en/de/es keys

release app 1.0.47 (code 48): Reconditioning RMA-only scan, A4 movement report, disposal option

Return label rebuilt: ONE label per return (was one per article and qty) rendered server-side by POST /api/print/return-label at the printer's real label size (registry → CUPS → legacy fallback, same as product labels) with size-driven layouts (tiny 35×17 / compact 54×29 / standard 74×52, 56×75, 102×152 with continuation pages), vector CODE128 of the RMA or customer-return document number, ad_org.companyname header band, partner · reason · locator line and an article table with quantities; mobile return page + desktop returns page now call the route instead of building jsPDF labels client-side

release app 1.0.48 (code 49): new return label (one per return, server-rendered)

Order sources: Amazon returns import flag (isImportReturns) + lookback days (importReturnsDaysFilter) wired through create/edit/view + store/update routes — toggle in the Marketplace-Specific tab's Amazon SP-API block gates the middleware's daily MFN returns → RMA import