Views overview

​

Known issues

​

Gaps in the current rules

• Whether the header’s second row is truly exclusive (a tab bar or a filter row, never both) needs confirming with the design team, not just read off a variant property. A variant property is how the design tool models a fixed set of choices — a header that could hold both would be built with two independent toggles instead, and which of those two structures is right can be a modelling decision rather than a design one. What ships either way: the frame has one Controls region below the header, so a page can currently show both. The page’s own constraint against this is the design’s apparent rule, and the question to ask is whether a page with sections and header filters was ruled out or just never built.
• The frame’s regions don’t map to the expected landmarks — already recorded above.
• The tab-bar workaround costs structure, not just appearance. Header puts every child into one row, so Tabs passed as a child would land beside Left and Right rather than beneath them — rendering it as a sibling instead reproduces the look but not the structure: the header no longer owns its bottom edge, and the bar arrives with no gutter — already recorded above.
• No archetype is fully documented. List view has its frame composition written and nothing else; the other six pages are stubs. Which frame regions each archetype requires, and what fills them, is undecided per archetype.
• No maximum content width. On a wide monitor nothing stops a full-width text body from running far past a readable line length — already recorded above.
• When a page gets a Footer is decided by Footers — unit of commit, view vs container ownership, fixed vs static. Still open there: partial validity under a view-level footer, what Cancel discards across several changed sections, and where a fixed footer sits on a narrow or short viewport.
• A left-aligned footer variant has no stated use case. justifyContent: "flex-start" pushes every footer action left. Nothing says when a page should reach for it, and it conflicts with the trailing-primary order stated elsewhere on this page.
• Pagination has no region. A paged list has nowhere in the frame to put its controls, and three unreconciled implementations exist — already recorded above.
• No step indicator exists. A multi-step flow can offer a backward action but has no way to show which step you are on or how many remain — already recorded above.
• When the leading slot applies is stated here and nowhere upstream. “A step in a sequence” is this page’s proposal.
• No maximum number of tabs. The design library’s tab component publishes slots for nine, and the page templates show five. Neither is a stated limit, and nothing says at what count a bar should have been a side rail or a set of pages.
• Whether a tab bar and Controls may both appear is settled in design and not in code. The page templates show a tabbed table with a filter row and a search field beneath the header, which is the order documented above. What the templates do not settle is whether those filters belong to the page or to the table — in the template they are drawn inside the data grid, not in a page region.
• Sub-tabs have a themed appearance and no rule. The rail’s indented mode exists in the theme, so a nested level is buildable. Nothing states when nesting is right, or how deep it may go.
• Tab addressability is stated here and nowhere upstream. Two approaches ship side by side — a bar of links, and a bar driven by local state with TabPanel. The page’s own constraint picks the first; that pick has not been signed off.
• Information has no content rule. It is a growth slot with no stated limit, which is how a header accumulates a paragraph.
• The site header is out of scope and undocumented. The frame stops at the page. What sits above it, and how a page coexists with it, is not covered anywhere. The boundary matters in one direction: Contents sizes itself by measuring the site header at runtime, so a page frame dropped into a shell that does not identify itself gets the wrong height — already recorded above.

​

Constraints Titan should enforce