family.py bump: a mount with a narrowed fetch refspec is refused member-to-local-only on a false detail, a commit only a tag holds is told "a pin the rest of the world cannot fetch", a mount with no origin is told to fix the network #686

Goal: where bump --to refuses a commit as unreachable from the member's remote, the refusal's reason is true of the mount in front of the person.

Filed by lane openRepoShape-5 on Brett Heap's word of 2026-10-16T12:17:02Z ("#459's family lines (~38)"), from lane 4's R613 lines on #459 (comment 6203972266, R613 p07, p03 and p10). Lane 4 judged each OPEN on 95780dd (cluster T8); re-read on d4a7fab.

Lines (line numbers at d4a7fab)

• R613 p07, bump fetches with plain git fetch -q origin, so a mount whose fetch refspec is narrowed keeps no remote-tracking ref for the member's other branches. scripts/family.py:5758 (cmd_bump); scripts/bump-leg.py:960-961 fetches +refs/heads/*:refs/remotes/origin/*. A commit fetched by hand from another branch is then refused member-to-local-only, with a false detail. family.py never makes such a mount (no --depth, --single-branch, shallow), so only one narrowed by hand reaches it. Not re-run; the code difference is read. Direction: fetch with the explicit refspec, as bump-leg.py does, keeping -q origin and the fetched = False path.
• R613 p03, a commit the member's remote holds only through a tag, which the mount also holds, is refused member-to-local-only with a reason that is untrue there. _checked_on_a_remote_branch (scripts/family.py:5329-5362) says "reachable from no branch of " and "A pin the rest of the world cannot fetch is a family the rest of the world cannot bootstrap" (:5360-5362), where a fresh clone holds the commit and a fetch by sha works. bump-leg.py's twin, bump-leg-commit-local-only ("somebody committed inside the leg's mount", :980), is untrue the same way. _checked_to_names_a_commit's docstring records that both tools ask about branches only and "Left as it is" (D2, option 1); this line is the wording, not the rule. Direction: keep the refusal and word the reason as what it is (no branch reaches the commit; where a tag of the remote holds it, say so, and that the way out is to push it to a branch), without the claim that nobody can fetch it. The bump-leg.py half is lane 5's Tracking: update-shape.py and bump-leg.py, keeping a project's shape copies and leg pins current #458 file: change both together or note the other.
• R613 p10, a mount with no remote named origin is told the stale-refs sentence and to "fix the network or the credential". cmd_bump's fetch of origin fails (scripts/family.py:5758, fetched = False), _on_a_remote_branch reads only refs/remotes/origin/ (:5305-5326), and _checked_on_a_remote_branch ends with the sentence naming the network (:5351-5354). Probed: in members/IRRS, git remote rename origin upstream, then bump --dry-run --to <HEAD~1> gives note: could not fetch members/IRRS from its origin and REFUSED member-to-local-only ... The fetch of members/IRRS FAILED ... fix the network or the credential. Direction: when the fetch fails, ask whether the mount has an origin at all (git remote get-url origin), and say it has none; the same sentence in bump-leg.py is Tracking: update-shape.py and bump-leg.py, keeping a project's shape copies and leg pins current #458's.

Left out of this issue, and why:

• R613 p08 (a fork as origin passes a commit only the fork has), p05 (a branch deleted after the last fetch still counts; pruning needs a ruling) and E8 (a holder pinned before family.py: bump --to refuses a commit only the member's mount holds, reachable from no branch of the member's remote, as bump-leg.py refuses its twin, so a holder is never pinned to a commit nobody else can fetch (#613) #625 to a commit only its mount held stays pinned; no validator asks) each need a decision of Brett's. They stay on Tracking: setup, scaffold, family and the shared helpers, creating projects and families #459.
• The tag-whose-commit-is-on-no-branch line (bump --to refuses it, add --at pins it) and R613 D1/p04 (the release commit's bump --to and the validator's own repair are refused) were accepted as ruled (D2, option 1, 2026-10-15T16:49:38Z) and recorded in _checked_to_names_a_commit's docstring (scripts/family.py:5365). D1's v1.1.0 release-notes sentence is owed to whoever writes docs/release-notes/v1.1.0.md (it is on README, AGENTS.md, docs/cli.md and the handbook: pipx install openreposhape is the second way in and the first on native Windows, git-filter-repo is installed in install.sh's order and never with a bare pip install, and a first draft of the v1.1.0 release notes (#563) #630's branch, not on main).

Overlaps: none with an open PR (#639 edits cmd_remove and remove's checks only). scripts/family.py and scripts/bump-leg.py are not pinned copies.

Mark the three lines → this issue on #459, and tick them there when this closes.

🤖 Generated with Claude Code