{# NO HEADING OF ITS OWN. This panel is the BODY of the card core builds for a loaded tool, and that card's header already carries an icon, the tool's name, the key that opens it, an eye and an X. A second name and a second close directly under them was one tool announcing itself twice in a 300px column -- and the two X's did different things, which is worse than redundant. Core owns the header; this owns what a card cannot know. Same bargain the transcripts layer strikes, for the same reason. See views/toolLoader.js and views/cardList.js. #}

{# The two ways this tool cannot work, kept apart because the fix differs. No metadata is a data-setup problem; nowhere to draw is a missing file. #} {# A mask was supplied and is still being converted. Cells here are drawn as the project's mask, and until the pyramid lands there is nothing to draw them on. Centroids would fill the gap, and that is what used to happen -- silently, and often for minutes, so the panel opened showing dots for a project whose whole point was cell shape, and nothing on screen said why or that it would change. This waits instead, and says how far along the job is. Anyone who would rather not wait can take the centroids deliberately; the mask still replaces them when it arrives. Fed by core's poll -- see main.js's plexora:segmentation-*. #} {# One line: what to colour by, and -- when that column is categorical -- the two controls that act on the list below it. No visible captions. The colour-by control is core's SearchableSelect in its button shape, so it reads as the column it is showing, and its dropdown carries the search that finds another one. The other two are a search field and a two-state All/None, which are what they look like. Each carries an aria-label, since there is no `for` to point a at a control that mounts its own generated elements. The category pair is hidden rather than moved for a continuous column: it belongs to the legend, and the legend is not there. #} {# Opens the ROI tool ALONGSIDE this one rather than in place of it -- see toolLoader's openToolAlongside. Regions are only worth hovering while the overlay they are summarised against is still drawn, so the usual single-active switch would take away the thing that makes the answer mean anything. A named button on a line of its own rather than another icon in the heading. Nothing about a polygon glyph says that hovering a region will break its cells down by the variable on screen, and an unlabelled control for a feature nobody knows exists is a feature nobody finds. The hint is the discoverable part; the button is just how it is reached. #}

Hover a region to see what it is made of.

{# Shown only for a column whose kind was genuinely a guess -- a numeric one with few whole values. See server/variables.py. #} {# Just the list. Its filter and its All/None live in the toolbar above, sharing that line with the colour-by control. #} {# The opacity slider used to be here. It is core's now, in the Cells section at the bottom of the sidebar, beside the mode buttons -- see index.html. Compositing is the viewer's, the same way geometry is, so every plugin that colours cells gets one slider rather than each shipping its own; and with several layers stacked, a slider inside one panel that moved whichever layer happened to be active was pointing at the wrong thing as often as not. This plugin still remembers the value per project: it listens for plexora:cell-layer-opacity-changed and saves the ones tagged with its own name. #}