Changes touching this path
- a shallow push asks whether a forge head is one it has not seen, as a complete push does since kkpruryt, and refuses before any object address leaves: refuse_unseen_forge_heads no longer returns early on a shallow position, and counts a forge head seen when it is held, retired, or reached by the declaration, whose reach keeps the frontier ids it names, so a forge head at the cut is not refused; a shallow position against a forge declaring its cut and a second writer root it holds neither of read both as a have of nothing and sent /wants, /stow and /ingest once each, stowing every change it held, before the forge refused it. a head strictly behind the cut is counted unseen, sound while no forge head is an ancestor of another, since the cut refusal has already asked that each frontier id be declared or reached from a held head; ADR 0098 says so. the shape with every forge head unheld and off the cut was already refused by that cut refusal before /wants. pinned: a_shallow_push_against_a_forge_head_it_does_not_hold_refuses_before_any_stow over the counting proxy, with a control namespace declaring the cut alone that takes the push. red before the change (0 passed and 1 failed) and under named mutations, each restored, failed and passed over it and the kkpruryt pin: the shallow early return put back (1 and 1, the refused push sent 1, 1 and 1), and the reach not consulted (1 and 1, 2 heads counted unseen for 1). cargo test green, 4953 passed over 145 test result lines with 15 ignored. owes no deploy (ntzsvkkv)
f15f2a25 · dbf3dbe6…
Renames are not followed. loot's tree maps a path to an address, so a rename is a delete and an add. This list is the history of the name, not of the bytes.