show progress arc while waiting for first map and clean up atomic state by JohnN193 · Pull Request #17 · viam-modules/cloudslam-wrapper

JohnN193 · 2026-03-20T19:46:27Z

Two related changes: a user-visible feature and the internal state refactor it motivated.

Progress arc
While a job is running but no map has arrived yet, PointCloudMap returns a point cloud arc instead of the blank default. The arc grows from a small sliver to a full circle over 5 minutes, giving visual feedback that the session is active. (Text label added in the next PR.)

Atomic state cleanup

Merge activeJob (job ID string) + jobStartTime into a single activeJobState atomic struct — both fields are now updated in one atomic swap, eliminating the window where one is set and the other isn't. Nil pointer replaces the empty-string sentinel.
Wrap updatingMapName / updatingMapVersion into an updatingMapInfo struct; nil replaces the != "" guard and fixes a latent bug where version could be set alongside an empty name.
Clear activeJob in StopJob to stop the polling loop, while preserving lastPointCloudURL so the final map stays visible after the session ends.
machine_part_id is now required at startup (matching machine_id behavior) — safe now that it's auto-populated from the env var.
Store logger on AppClient at construction; removes the logger parameter from CheckSensorsDataCapture.

Fix

Add page=slam to the StopJob return URL (without it the app backend redirects to /fleet/location/...).

Part of a series: [PR A] → [PR B] → PR C → [PR D]

JohnN193 · 2026-03-20T21:02:23Z

cloudslam/cloudslam.go

+// generateProgressRingPCD generates a point cloud of a progress arc indicating elapsed time.
+// The arc grows clockwise from 0 to a full circle over progressRingDuration, giving the user
+// visual feedback while waiting for the first cloudslam map to appear.
+func generateProgressRingPCD(elapsed time.Duration) ([]byte, error) {


when running cloudslam there is always a period of waiting for the job to start, because we have to spin up a job and setup the machine.

now we display something that moves, and in the next pr we expand on that further

nfranczak · 2026-03-23T14:22:51Z

cloudslam/cloudslam.go

+
+	pc := pointcloud.NewBasicEmpty()
+	for i := 0; i < filledPoints; i++ {
+		angle := float64(i) / numPoints * 2 * math.Pi


aren't you duplicating work here?

kk going to update the code to store the pointcloud in state rather than recreate everytime

nfranczak · 2026-03-23T14:31:57Z

cloudslam/cloudslam.go

+	}
+
+	var buf bytes.Buffer
+	if err := pointcloud.ToPCD(pc, &buf, pointcloud.PCDAscii); err != nil {


ascii is large. could do binary

While a job is active but no map has been received yet, PointCloudMap returns a PCD point arc instead of the blank default. The arc grows from a small sliver to a full circle over 5 minutes, giving the user visual feedback that the session is running. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

- merge activeJob + jobStartTime into activeJobState struct; nil pointer replaces empty-string sentinel, and startedAt replaces the separate jobStartTime atomic so both fields are updated atomically - wrap updatingMapName/updatingMapVersion into updatingMapInfo struct; nil replaces the != "" guard and fixes a latent bug where version could be set with an empty name - clear activeJob in StopJob (stops polling) while keeping lastPointCloudURL so the final map stays visible after a session ends - store logger on AppClient at construction; drop logger param from CheckSensorsDataCapture Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

part_id is now required (from config or env var), matching the existing machine_id behavior. removes the dead partID != "" guard in initialize. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

- ParseSensorsForPackage now delegates to sensorInfoToProto instead of re-iterating svc.sensors - generateProgressRingPCD uses PCDBinary instead of PCDAscii for smaller output Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

- Background thread now owns the current PCD entirely: builds the progress arc incrementally (appending only new points per tick) while waiting for a real map, then downloads and stores map bytes once available. PointCloudMap just reads the stored bytes. - Removes lastPointCloudURL and progressPCD atomics in favor of a single currentPCD atomic.Pointer[[]byte] - ParseSensorsForPackage delegates to sensorInfoToProto to avoid duplicating sensor iteration - generateProgressRingPCD uses PCDBinary instead of PCDAscii Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

JohnN193 · 2026-03-23T15:26:54Z

cloudslam/cloudslam.go

-		mapURL := resp.GetMapUrl()
-		svc.lastPointCloudURL.Store(&mapURL)
+
+		if mapURL := resp.GetMapUrl(); mapURL != "" {


moved all of the pcd generation logic to the background thread, so now we don't have multiple threads making pcds

the background thread builds them, and GetMap sends them

nfranczak · 2026-03-23T16:17:42Z

cloudslam/cloudslam.go

+			}
+		} else if !job.startedAt.IsZero() {
+			// no map yet — incrementally build the progress arc
+			fraction := math.Max(0.02, math.Min(time.Since(job.startedAt).Seconds()/arcProgressRingDuration.Seconds(), 1.0))


0.02 so we always start with some % of the circle showing up

nfranczak · 2026-03-23T17:19:07Z

cloudslam/cloudslam.go

+			}
+		} else if !job.startedAt.IsZero() {
+			// no map yet — incrementally build the progress arc
+			fraction := math.Max(minArcPercent, math.Min(time.Since(job.startedAt).Seconds()/arcProgressRingDuration.Seconds(), 1.0))


cloudslam/cloudslam.go

nfranczak · 2026-03-23T17:23:00Z

cloudslam/cloudslam.go

+					angle := float64(i) / arcNumPoints * 2 * math.Pi
+					x := arcRadius * math.Cos(angle)
+					y := arcRadius * math.Sin(angle)
+					if err := arcPC.Set(r3.Vector{X: x, Y: y, Z: 0}, pointcloud.NewBasicData()); err != nil {


opt: 0 should be a const

nfranczak

just the optional to have

arcPC.Set(r3.Vector{X: x, Y: y, Z: 0}

become

arcPC.Set(r3.Vector{X: x, Y: y, Z: someConstName}

JohnN193 mentioned this pull request Mar 20, 2026

2026 cloudslam updates #14

Closed

JohnN193 force-pushed the pr-b-config branch from a4302c1 to 6d21e85 Compare March 20, 2026 19:52

JohnN193 force-pushed the pr-c-state-cleanup branch 2 times, most recently from c2df93b to 71b5798 Compare March 20, 2026 19:58

JohnN193 force-pushed the pr-b-config branch 2 times, most recently from 6a13444 to de03d7f Compare March 20, 2026 20:08

JohnN193 force-pushed the pr-c-state-cleanup branch from 71b5798 to 151372c Compare March 20, 2026 20:08

JohnN193 force-pushed the pr-b-config branch from de03d7f to 79f5cc3 Compare March 20, 2026 20:16

JohnN193 force-pushed the pr-c-state-cleanup branch from 151372c to 6d5fc3d Compare March 20, 2026 20:16

JohnN193 commented Mar 20, 2026

View reviewed changes

nfranczak reviewed Mar 23, 2026

View reviewed changes

JohnN193 force-pushed the pr-b-config branch from 79f5cc3 to d892037 Compare March 23, 2026 14:26

nfranczak reviewed Mar 23, 2026

View reviewed changes

Base automatically changed from pr-b-config to main March 23, 2026 14:40

JohnN193 and others added 6 commits March 23, 2026 10:41

fix url

fd1e8f9

tweak comments

1495303

claude self review

cfa0a08

require machine_part_id at startup

3280815

part_id is now required (from config or env var), matching the existing machine_id behavior. removes the dead partID != "" guard in initialize. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

JohnN193 force-pushed the pr-c-state-cleanup branch from 6d5fc3d to 3280815 Compare March 23, 2026 14:42

JohnN193 and others added 2 commits March 23, 2026 10:53

JohnN193 commented Mar 23, 2026

View reviewed changes

JohnN193 requested a review from nfranczak March 23, 2026 15:27

nfranczak reviewed Mar 23, 2026

View reviewed changes

add const

dc4d10b

nfranczak reviewed Mar 23, 2026

View reviewed changes

cloudslam/cloudslam.go Show resolved Hide resolved

nfranczak reviewed Mar 23, 2026

View reviewed changes

make 100 percent a const

bf28460

nfranczak approved these changes Mar 23, 2026

View reviewed changes

JohnN193 merged commit cb67565 into main Mar 23, 2026
1 check passed

JohnN193 deleted the pr-c-state-cleanup branch March 23, 2026 18:16

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

show progress arc while waiting for first map and clean up atomic state#17

show progress arc while waiting for first map and clean up atomic state#17
JohnN193 merged 10 commits intomainfrom
pr-c-state-cleanup

JohnN193 commented Mar 20, 2026

Uh oh!

JohnN193 Mar 20, 2026

Uh oh!

nfranczak Mar 23, 2026

Uh oh!

JohnN193 Mar 23, 2026

Uh oh!

nfranczak Mar 23, 2026

Uh oh!

JohnN193 Mar 23, 2026

Uh oh!

nfranczak Mar 23, 2026

Uh oh!

JohnN193 Mar 23, 2026

Uh oh!

nfranczak Mar 23, 2026

Uh oh!

Uh oh!

nfranczak Mar 23, 2026

Uh oh!

nfranczak left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

JohnN193 commented Mar 20, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

nfranczak left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants