CLAUDE.md's Caveats section states:
Re-popping the same panel isn't possible (it's removed from the layout on the first pop)
The parenthetical doesn't support the claim as a general statement. "Removed from the layout" describes the mechanism of popping, not a barrier to repetition. The same document's Pop-out section describes dockFollowerPanel() → _redockPanel() re-instating a docked panel back into the main layout — after which nothing described prevents popOutPanel() being called on it again.
Two readings are possible and the doc should pick one:
- "A panel can't be popped twice concurrently" — trivially true of any panel, and not really a caveat worth calling out.
- "Once popped, even after being docked back, that panel can never be popped again for the rest of the session" — if this is actually true, the real blocking mechanism (e.g. a
popupId that's consumed/exhausted and not reset on dock) needs to be named, since nothing in the documented dock flow explains why it would hold.
Suggest verifying against the actual behavior (pop → dock → pop again) and either removing the caveat (if reading 1 is what was meant and it's not actually noteworthy) or naming the real mechanism (if reading 2 is accurate) and possibly filing it as a bug if unintentional.
CLAUDE.md's Caveats section states:
The parenthetical doesn't support the claim as a general statement. "Removed from the layout" describes the mechanism of popping, not a barrier to repetition. The same document's Pop-out section describes
dockFollowerPanel()→_redockPanel()re-instating a docked panel back into the main layout — after which nothing described preventspopOutPanel()being called on it again.Two readings are possible and the doc should pick one:
popupIdthat's consumed/exhausted and not reset on dock) needs to be named, since nothing in the documented dock flow explains why it would hold.Suggest verifying against the actual behavior (pop → dock → pop again) and either removing the caveat (if reading 1 is what was meant and it's not actually noteworthy) or naming the real mechanism (if reading 2 is accurate) and possibly filing it as a bug if unintentional.