family.py: follow-ups from RV1117's review of #1117 (#1005): the add that member-unknown names is refused on the tracking branch, the bare family.py, the <org> blank, and four older ones #1127

(lane openRepoShape-4, 2026-10-21T15:09:47Z)

These are seven follow-ups from RV1117's independent review of PR #1117 (#1005) at 10dc011c (#1117 issuecomment-6318620190). None of them blocks #1117, which RV1117 found READY. N1 to N3 are about the three lines #1117 adds; N4 to N7 were there before it.

1. N1: on the holder's tracking branch, the add that bump's member-unknown names is refused. A fresh clone stands on the tracking branch, and there family.py add --family-root <holder> --member <org>/<P> is refused family-on-tracking-branch.
   • That refusal prints its own git switch -c line, so the chain still ends in a command that works. But the remedy reads as if it runs as printed.
   • The two remove lines can be reached there only from --dry-run.
   • Fix: say in member-unknown's remedy that the add runs off the tracking branch, or print the git switch -c line first.
2. N2: the three lines start with a bare family.py, as family.py and bump-leg.py: every remedy that names family.py bump names --family-root, so each runs as printed (#927) #1002's bump lines do. _add_command and validate-family.py print the interpreter and the script's path instead.
   • Fix: print every remedy's command the way _add_command does, from one helper.
3. N3: <org> is left as a blank, although family.yaml's org: is known. Pasted unfilled, <org>/IRQS is a redirection in every shell, not a value. Where a file named org is in the working directory, it tries to create /IRQS, which an ordinary user is refused.
   • Fix: fill in the org from family.yaml where it is set, and leave the blank only where it is not.
4. N4 (older than family.py: the remove that both member-duplicate-id remedies name and the add that bump's member-unknown names carry --family-root <holder>, quoted for the host, so each runs as printed (#1005) #1117): the row's --member name is printed raw, where validate-family.py quotes it. It changes no outcome today.
   • Fix: quote it with shell_quoted, as the holder now is.
5. N5 (older than family.py: the remove that both member-duplicate-id remedies name and the add that bump's member-unknown names carry --family-root <holder>, quoted for the host, so each runs as printed (#1005) #1117): nothing ties the README sentence's list of verbs to NAMED in tests/test_readme_windows_quoting_sentence.py. RV1117 found this by reading the test, not by running it.
   • Fix: have the test check the sentence's verbs against NAMED.
6. N6 (older than family.py: the remove that both member-duplicate-id remedies name and the add that bump's member-unknown names carry --family-root <holder>, quoted for the host, so each runs as printed (#1005) #1117): a holder path that holds a backtick makes the backtick-wrapped remedy hard to read. The shells still run it correctly.
   • Fix: leave out the backticks around a command that holds one.
7. N7 (older than family.py: the remove that both member-duplicate-id remedies name and the add that bump's member-unknown names carry --family-root <holder>, quoted for the host, so each runs as printed (#1005) #1117): member-already-present says "remove takes it out" without a line to type, as W1005's PR notes already say.
   • Fix: print the remove line with --family-root, as the other remedies now do.

Lane: openreposhape-4 (openRepoShape-4)