Make publication time mean something

Saturday’s work gave Steve’s publishing tools a clearer contract: creating a draft, scheduling it and publishing it are separate events. That distinction reached the database, worker and interface. Delivery status also mattered: DOM Studio reached npm, the framework passed CI, and Scout’s release was prepared without evidence of promotion.

• GrowthScout · separate generation from scheduled publication. Publish now and Publish on… gained separate paths, with rescheduling, cancellation and a default 09:00 target in the website’s timezone. Generation saves drafts; an every-minute worker selects indexed batches of due articles, with concurrency protection and bounded retries. Multiple articles can share a day. Focused tests, types and conventions passed. The changes were subsequently committed and pushed, and Docker release preparation completed. Production promotion and migration application were not established; existing overdue automatic drafts still need review before activation.
• GrowthScout and db3 · keep UTC instants separate from local intent. Website setup gained editable timezone suggestions from audience location. Framework database connections and timestamp conversion explicitly enforce UTC, while named IANA zones remain separate for display and scheduling. Temporal sits behind one internal polyfill module so application APIs continue returning Dates and ISO strings. The initial foundation passed 975 framework tests and 12 focused Scout tests; later container checks verified timezone behaviour. Scout’s Docker runtime was aligned with Node 24. The framework changes reached the tagged beta.4 revision and passed GitHub CI, but npm publication remained awaiting authentication at Saturday’s cutoff.
• DOM Studio · publish the selected-time visibility repair. Opening the datetime picker now reveals the selected time after the popover has laid out. The control compares the option’s offsets against the scrolling list and adjusts only that list, avoiding a page-level scroll. Version 0.2.6 was published directly from the validated local package; registry integrity and the latest tag were verified, and the GitHub Release followed. Hosted website deployment and release-history synchronization remained blocked by GitHub Actions billing. Scout’s later pushed change also updated its DOM Studio dependency.
• GrowthScout and DOM Studio · identify the misleading blog date. The live destination record showed its publication date copied from draft creation, while Scout’s embedded publication timestamp and the destination update matched the following day’s publishing run. Source inspection located the receiver assignment. This explained the displayed date without proving an early publication. The investigation also identified the exact-day lookup’s overdue-draft gap, which informed the new due-time worker. Repairing the receiver and historical dates was recommended; completion of that separate repair was not established.
• GrowthScout · restore the article editor and simplify planner labels. The editor resolved an older Tiptap core missing a required export. A compatible dependency and refreshed Vite bundle restored the local editor; five editor tests, types, conventions and the production build passed. Planner cards now show a compact planned time in the website’s timezone. A separate layout investigation measured two cards each consuming an entire day cell’s height, explaining their overlap into later weeks. That measurement established the cause; it should not be mistaken for proof that the overlapping-card repair shipped.
• GrowthScout · let full-width screens opt out of layout padding. AppLayoutSidebar gained a contentPadding prop that defaults to true. The article page opts out explicitly, letting the parent own spacing without child margin workarounds. All 20 sidebar tests, types and conventions passed, with desktop and mobile review. A separate mobile toolbar clipping issue remained. The initial scheduling interface also exposed a verification gap: isolated interaction checks missed its hierarchy and density inside the real screens. The project guide now requires review in the actual app with realistic content before declaring UI work complete.
• GrowthScout · distinguish prerequisite waits from exhausted retries. A local article-generation check found a 15-second deferral while preparation completed. Those waits do not spend the generation retry allowance. Original database evidence then showed preparation complete and the article worker processing the job, with the image worker running. No restart or manual retry was needed. Processing was confirmed; final article completion was not.
• DOM Studio and GrowthScout · define evidence and ownership for new data features. Grid research proposed a database-independent client datasource for loading and caching, with a separate server SQL adapter for configured tables, validated fields and tenant permissions. This remains a proposed architecture, not an implemented database connector. A separate AI-visibility proposal paired prompts with real Google result queries, recording location and device; an AI-generated estimate would not establish search rank. Neither research thread demonstrated implementation.