Improvement: When updating all in stack view - last queue item shifts to the ungrouped stack

[begunfx]

When we try to update a full stack of containers using the "Update All" in stack view I noticed a peculiar behavior. When the last container in the stack to update starts updating it and the group that I triggered to update disappear with the final updating container reappearing in the ungrouped stack. From a UX perspective when a full stack/group is triggered to update only after the last container has completed the update, then the stack/group should disappear so the container update process is more consistent.

[scttbnsn]

Thanks for the detailed report, @begunfx — and good catch.

You're right that a stack should stay intact for the whole update rather than dissolving the moment its last container starts. Here's what was happening: when a container updates, Drydock briefly removes the old container before the recreated one appears. For the last container in a stack, that left the stack with only a single live member for a split second — which tripped the rule that flattens genuine single-container stacks into the Ungrouped section (the behaviour you requested in #179). So the stack header vanished and the remaining container jumped to Ungrouped.

The fix makes the flatten decision key off how many containers are assigned to a stack, not how many happen to be live at that instant. A multi-container stack now stays grouped throughout the entire "Update All" run, even while its last container is mid-recreate. Genuine single-container stacks still flatten as before.

This will land in the next v1.5 release candidate.

[scttbnsn]

@begunfx this one's fixed and shipped.

A multi-container stack now stays grouped for the whole "Update All" run instead of dissolving when the last container starts updating. What was going on: when a container updates, the old one is removed for a moment before the recreated one shows up, and for the last container in a stack that briefly left the stack with a single live member, which tripped the rule that flattens genuine single-container stacks into Ungrouped (the #179 behavior). The fix keys the flatten decision off how many containers are assigned to the stack, not how many happen to be live at that instant, so a real multi-container stack holds together through the recreate. Genuine single-container stacks still flatten like before.

It landed in v1.5.0-rc.25 and is in v1.5.0 GA and the current v1.5.1-rc.1. Update and run an "Update All" on a stack and the group should stay intact the whole time now. Let me know if you still see it bounce to Ungrouped on your setup and I'll dig back in.