life: let a row retire from the ties it declared #165
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "land/cap/liveness"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Ending a row scanned the whole model for anything still pointing at it and
refused if it found one. Half of that scan looked at the row's own bonds: a
team could not retire while it still had members, a room could not close while
it still listed guests. A bond has no lifecycle of its own — the model says so
where it draws the line between a bond and a resource — so a row was being held
open by relations it had declared itself.
The cost was not inconvenience. Dissolving a group meant unpicking its members
one at a time, and each cut dropped one person's authority, so the estate spent
the whole operation inside a half-dissolved group. Revoking authority should be
one write.
It now is, because the other half of this change makes a group principal count
only while its own row is live.
bearsresolvedwho = "team 5"by reading themembership ties and never asked whether team five still existed; with the veto
gone that would have left a grant working through a retired principal. The two
halves are one statement and land together.
A tie pointing at the row from elsewhere still refuses the end. That direction
is a separate question about how liveness should descend a containment chain,
and it is not answered here.
The tests that asserted the old law now assert the new one: a follower may
retire while its own follow lives, an enrolment does not hold its student open,
and a dissolved team stops granting while its membership rows are still there.
Co-Authored-By: Claude Opus 5 (1M context) noreply@anthropic.com