You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The insert marker which appears when session are reordered is drawn correctly when a session is moved upwards but it is misleading/wrong when a session is moved downwards. Why? Because it marks the same position (above the element). A simple fix could be to leave the marker as it is when a session is moved upwards but to put the marker to the bottom of a session when it is moved downwards. The gif animation shows both scenarios.
The first run demonstrates the incorrect behavior then the session is moved to its original position and then it shows the correct behavior but with bad ux in the second half of the gif.
Fixed in v0.38.0. Your diagnosis was exactly right, and the gif made it obvious — thanks for taking the time to record both halves.
What was happening. The drop handler removed the dragged row and re-inserted it at the hovered row's index. Dragging up, that lands above the target. Dragging down, removing an earlier element shifts every later index by one, so the same code lands it below. The marker was drawn above the hovered row either way, so it described one direction correctly and the other not at all.
On your suggested fix. Direction-dependent markers would have matched the behaviour, but you spotted the cost yourself in the second half of the gif: you have to aim one row past where you want the session to end up, and the marker's meaning changes depending on which way you came from. So I went one step further and removed direction from the equation entirely — the pointer's half of the row picks the edge:
pointer in the top half → marker on the top edge → lands before that session
pointer in the bottom half → marker on the bottom edge → lands after it
Marker and landing spot now come from a single value, so they cannot disagree. Dropping onto an edge a session already sits on is a no-op rather than a pointless re-render, the marker is drawn inset so the bottom edge of the last row isn't clipped, and it clears when you drag out of the list instead of lingering on the last row you passed.
Covered by tests including both directions × both edges, the list ends, no-ops, and a check that no reorder ever drops or duplicates a session.
Activity
fix: browser refs, drop marker, in-app update, dead popup links (#123 #…
amirlehmam commentedon Jul 28, 2026
Fixed in v0.38.0. Your diagnosis was exactly right, and the gif made it obvious — thanks for taking the time to record both halves.
What was happening. The drop handler removed the dragged row and re-inserted it at the hovered row's index. Dragging up, that lands above the target. Dragging down, removing an earlier element shifts every later index by one, so the same code lands it below. The marker was drawn above the hovered row either way, so it described one direction correctly and the other not at all.
On your suggested fix. Direction-dependent markers would have matched the behaviour, but you spotted the cost yourself in the second half of the gif: you have to aim one row past where you want the session to end up, and the marker's meaning changes depending on which way you came from. So I went one step further and removed direction from the equation entirely — the pointer's half of the row picks the edge:
Marker and landing spot now come from a single value, so they cannot disagree. Dropping onto an edge a session already sits on is a no-op rather than a pointless re-render, the marker is drawn inset so the bottom edge of the last row isn't clipped, and it clears when you drag out of the list instead of lingering on the last row you passed.
Covered by tests including both directions × both edges, the list ends, no-ops, and a check that no reorder ever drops or duplicates a session.