Dashboard components

What a dashboard component is: the catalog of kinds, the addresses they bind, how a reading is drawn, and the property sections every kind shares.

A component is one item on a dashboard: a rectangle of grid cells that draws something. Most components read the live address space and present it, a few write to it, and a few only draw. You place one from the Add area of the dashboard editor's tool rail, by dragging it onto a precise cell or clicking it to take the first free one, and it opens in Properties for you to configure.

Every kind has a reference page of its own, listed under The catalog and in this group's own list. This page holds what they all have in common: the addresses they bind, how a reading is drawn, and the property sections that repeat on every kind. Which role may press a control, and the lock mark a component wears when the signed-in role may not, are on View and dashboards.

The catalog

The Add palette groups the catalog by category, and the categories say what a family does. Display is everything that reports: the plain readouts and tables, the lamps, and the charts that follow points over time or compare them side by side. Control is the family dedicated to writing: every one of its components exists to put a value on a point. It is not the only surface an operator presses, though. Alarm brings the pending registry onto a screen, and its banner is acknowledged from there, per row and for the lot; the Procedure cards listed below are pressed too. Media places pictures, library marks and industrial artwork. Layout groups a region behind the rest of the screen.

Beyond the palette, the generated unit panels carry the Procedure cards, placed by the Process workspace and not offered in Add: Interlock, Procedure picker, Occurrence list and Comment timeline.

The pictures, marks and artwork the Media kinds draw come from the station's own library on the Logic page: The asset library holds the images, icons and colors, and Symbols the catalog a Symbol is picked from, with the terminals its connections are drawn between.

Addresses and live readings

Live components bind to addresses: rooted names into the station's live address space, the same space Logic reads (see the Logic page). An address names one point, wherever it lives:

Address fields are free text with suggestions: the list is drawn from the live point list, capped at 50 rows, narrowed to the value category the field consumes (a gauge suggests numbers, a toggle leads with Bool points, a write control lists only what can be written), and every row is marked with its type. You may always type an address the list does not offer, including one that does not exist yet. Once a field names a point, the editor shows what that point reads right now under the field: the confirmation a component drawing no number of its own cannot give. On a screen defined on a Process model, the model's abstract channels lead every list; picking one binds the channel and leaves the unit to be answered when a unit opens the screen.

The canvas refreshes its components from the address space every 150 ms on the station's own window and every 400 ms on a remote browser, over the values the logic sweep publishes. Two rules hold for every live component:

• Never a stale value. A point that stops reading drops to that component's own placeholder (the gap dash, a hollow lamp, a chart gap), never the last number.
• One bad-reading mark. A component whose bound point fails to read (the address resolves to nothing, or the point answers with bad quality) wears one shared mark: a dashed caution outline around the cell and a small circled ! badge in its top right corner. Every kind wears the same mark, so it is learned once. A resolved point answering good quality with no value is not a failure (a bench between runs, a counter outside a cycle), so the component keeps its quiet placeholder and the mark stays dark. An unbound address is nobody's fault: a screen still being assembled never reads as a screen of warnings.

Value display

How big a reading is drawn is governed by one contract, shared by every text surface on the canvas (readouts, captions, key faces, state words):

• Every surface is either automatic (the cell decides the size, bounded by the width the surface's declarations reserve: digits, decimals, unit, or the authored words) or stated, a size in points you give (9 to 144). A stated size is literal: it renders at that size in any cell and is never reduced to make content fit. Nothing about the size ever follows the value that happens to arrive, so a number ticking between four and six characters never moves the type.
• A number is never drawn cut. Where it cannot be shown whole (more digits than the field declares, or a stated size wider than the cell) the surface draws ####, the spreadsheet's and the DCS's mark for a field too narrow for its value, with the whole reading in the tooltip.
• Free text truncates with an ellipsis, whole text in the tooltip. Words never wear the number's mark.
• A point that is not reading is a different fact again: the gap dash, plus the shared bad-reading mark where the silence is real failure. Absence of a reading and absence of room never borrow each other's mark.
• Numbers rest on the right of their field, in tabular figures, so columns of readings align with themselves.

The 11 px readability floor holds everywhere, and the shared Content scale knob multiplies the resolved size: down always shows (still floored), up grows only as far as the line still fits uncut.

The shared property sections

Selecting a component opens its properties. The sections named for the component's own kind come first (its content, its data, its format) and are documented on that component's page. Around them, every component shares the same sections. The panel that draws them, the live reading under every address field, and the controls that repeat across kinds (display size, Reads as, the Scale bands, the chart toolbox) are described on The dashboard editor.

Identification names the component for whoever edits the screen: the name heads the property panel, labels the row in the Components list and answers its search. Nothing draws it on the operator's screen; that is the Appearance title.

Appearance is the component's box:

Position and layer holds the numeric X / Y / W / H, the layer (Z) and the Bring to front / Send to back pair. Advanced appears only on the kinds that have infrequent extras. Actions duplicates or removes the component.

Hold Ctrl to select several components and edit what they share: the Appearance fields fan out across the whole selection (a field where members disagree says "Mixed"), the layer commands keep the batch's relative order, and Duplicate/Remove act on all of them. Per-kind fields need a single component.