Add: Data from diodes, readback/setpoint, plot tools, scan update by tomkane-dls · Pull Request #235 · DiamondLightSource/atlas

tomkane-dls · 2026-10-16T12:27:51Z

No description provided.

…a, and allow switching

The poll that should fetch a scan's last points once it finished never ran: it was conditional on the poll interval still being set, but the previous run of the effect had already cleared it. So the final image was whatever the last regular poll got, up to half a second before the scan ended. It now runs whenever a scan stops running. /binned answers 503 with Retry-After while the data file is behind the catalog. That was logged as a polling error; it is now treated as "not ready yet": during a scan the next poll simply tries again, and the final poll, which nothing follows, retries after the time the server asks for (up to 10 times). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BsMDFhupHcBFbVJPK8QAvR

A quiet line at the bottom right of the plots says where the plotted scan is: "Scanning…", then "Scan complete", which fades out after a few seconds. While the final poll waits on a 503 it says the last points are loading, and if they never arrive, or the scan failed, it says so in the warning or error colour and stays. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BsMDFhupHcBFbVJPK8QAvR

…dges Cells no point had landed in yet were drawn black, the same as a measured low value, so a scan in progress looked like dark data. They are now faint diagonal stripes, one in the colour of the page behind the plots and one a shade darker on a light page or lighter on a dark one, so they recede in either theme. Each cell is drawn as 4x4 pixels to carry the pattern. A run without diodes shows its diode plots as all stripes, the camera's shape. A flat image (a single point, or all values equal) was scaled to black, since its range is zero; it is now drawn at full brightness. The plots were drawn in pixel numbers, without axes. They now get /binned's bin edges as axes, so each cell sits between its own edges in scan coordinates. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BsMDFhupHcBFbVJPK8QAvR

…n or footprints A Plot Type choice under Data Source picks the drawing, with a tooltip on what each shows. Setpoint grid is /binned's grid as before; the others come from Tiled's /render, drawn over the scan's extent at 200 px on its longer side: a step-sized rectangle at each recorded position (Readbacks), a beam-sized Gaussian on each point, or each point's beam footprint with overlaps split between neighbours. Both data sources work in every drawing. Where /render drew nothing, or covered a pixel only partly, it fades into the empty-cell stripes, so gaps the beam left stay visible. Gaussian and footprints use the beam size the run records. No run records one yet, so /render falls back to one scan step, and a note under the plots says so, with the size used. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BsMDFhupHcBFbVJPK8QAvR

The drawing is now two choices of one line each. Positions: Setpoints (the scan's grid, each cell the mean of the points read back in it) or Readbacks (each point where the motors read back). Plot Type: Rectangles, Gaussian or Footprints. Gaussian and footprints only make sense at the readback positions, so on the setpoint grid they are greyed out and the grid's cells show as rectangles; the shape picked for readbacks is kept meanwhile. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BsMDFhupHcBFbVJPK8QAvR

The empty cells' 2-pixel stripes were hard to tell from dark data in the dark theme. They are now the page's background crossed by thin diagonal lines, lighter than a dark background and darker than a light one. For lines that thin, the setpoint grid is drawn at least 240 pixels on its longer side (and at least 4 per cell), with the axes split to match. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BsMDFhupHcBFbVJPK8QAvR

hyperrealist

This should work as is, and it is interoperable with the now deployed dls-tiled 0.4.0. But please review #238 and merge it to this branch first before you merge this.

claude:

What was checked, and small notes (most addressed in #238)

Checked

Merged with current main (incl. the tomography threshold change), it merges cleanly and type-checks, tests, lints and builds.
Against dls-tiled 0.4.0 (deployed on b01-1), /binned returns DiodesRed/DiodesGreen/DiodesBlue for runs with diodes and null for runs without, which this handles.

Notes

The coverage check fails: ViSR coverage drops 1.08% → 1.03% because the new code has no tests. #238 adds tests for it.
DiodesRed/DiodesGreen/DiodesBlue are typed number[][], but are null for a run without diodes. It works at runtime (toRgbNdt falls back to the placeholder), but the type should allow null. Fixed in #238.
With Diodes selected, a run without diodes shows the small black placeholder, which reads as dark data. #238 draws it as empty cells instead.
The MSW mocks have no Diodes* data, so the Diodes view is empty in mock mode.
SliceViewer now keeps its own copy of enum Plane, which has to stay in the same order as plane.tsx.

ViSR: plot types, final poll and scan status on top of the Enz Locum UI

ViSR: create shared state hook, apply to all controls

tomkane-dls · 2026-11-04T14:56:07Z

I've merged in a PR that repurposes abigail's code allowing sliceviewer's controls to persist across tabs, and applying it to the new controls here, as it was tightly coupled with whats on this one

Make the Positions and Plot Type help text user-facing (no "spec" or "setpoint grid" jargon), add tooltips to the Data Source options, and say "beam footprint" rather than "beamline footprint" in the beam size warning. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

hyperrealist · 2026-11-04T17:17:15Z

@EmsArnold could you take another look at the tooltips?

EmsArnold · 2026-11-05T07:53:52Z

Tooltip changes look great, thanks!

The only other thing I noticed was that you can make the dividing line cut through buttons (on both the spectroscopy page and the tomography page) if you make the width of your browser sufficiently small. I'm not sure this is an issue given we always aim to have the same size window, but may be something which needs to be looked at in the future.

tomkane-dls · 2026-11-05T14:08:42Z

Tooltip changes look great, thanks!

The only other thing I noticed was that you can make the dividing line cut through buttons (on both the spectroscopy page and the tomography page) if you make the width of your browser sufficiently small. I'm not sure this is an issue given we always aim to have the same size window, but may be something which needs to be looked at in the future.

Yes Abigail and I noticed this, it's now visible on this PR because we added controls for spectroscopy, but it was already introduced for tomography on a previous branch. Once things have settled I will try and come up with a good solution for this

reimplement some of #221 - change response api to fetch enz locum dat…

8fec504

…a, and allow switching

tomkane-dls mentioned this pull request Oct 16, 2026

Placeholder fetch, plots, and controls for enz locum and composite data #221

Closed

tomkane-dls requested a review from hyperrealist October 16, 2026 12:28

hyperrealist mentioned this pull request Oct 19, 2026

Bin the RGB diodes beside the camera in /binned DiamondLightSource/dls-tiled#24

Merged

hyperrealist and others added 6 commits October 19, 2026 14:14

hyperrealist mentioned this pull request Oct 19, 2026

ViSR: plot types, final poll and scan status on top of the Enz Locum UI #238

Merged

hyperrealist approved these changes Oct 19, 2026

View reviewed changes

Merge pull request #238 from hyperrealist/visr-235-features

f90a8fa

ViSR: plot types, final poll and scan status on top of the Enz Locum UI

tomkane-dls changed the title ~~Fetch Enz locum data and allow switching between datasets~~ Add: Data from diodes, readback/setpoint, plot tools, scan update Oct 20, 2026

tomkane-dls marked this pull request as ready for review October 20, 2026 09:11

Merge remote-tracking branch 'origin/main' into enz-locum

f9dda9e

tomkane-dls requested review from EmsArnold and douglaswinter October 21, 2026 14:44

tomkane-dls assigned tomkane-dls and hyperrealist and unassigned tomkane-dls Oct 21, 2026

create shared state hook, apply to all controls

251f245

tomkane-dls mentioned this pull request Oct 22, 2026

Improved scan status #249

Draft