shape-doctor.py and siblings.py: a .gitmodules section git ignores ([submodule ""], a name with a .. component, a url beginning with -) is still read as a member, where git says No url found and exits 128

Goal: the doctor and make siblings read a holder's .gitmodules as git does, and skip what git skips. Then neither one names a mount url, a NEXT line or a clone for a section that git will not mount.

Filed by lane openRepoShape-3 under v1.1.1, on Brett Heap's word of 2026-10-16T13:35:28Z ("yes create v1.1.1"). Lane 3's reviewer RV-c found it while reviewing #700 (#668), as S3. Nothing here regressed: main, #700 and siblings.py all read these sections, and the two tools agree with each other.

What happens. RV-c checked each case against git submodule init -- <path>, in a fresh repository with a real gitlink.

git ignores or refuses these:

[submodule ""]: git ignores it ("ignoring suspicious submodule name"). The doctor and siblings.py read it as the name ''.
A name with a .. component, such as ../evil: git's check_submodule_name refuses it.
A url beginning with -: git's looks_like_command_line_option refuses it.

For each of the three, git says No url found for submodule path … and exits 128.

Two more cases:

A url key with no value: git refuses the whole file (bad config line 3).
A section with a url and no path: both tools key it by its name, but git mounts it nowhere.

What to do:

In the doctor's .gitmodules reading (submodule_config, gitmodules_entry) and in siblings.py's, drop exactly the sections git drops, by git's own tests: an empty name, a name with a .. component, and a url beginning with -.
Read a file that git refuses whole as having no entries.
Pin every case in the parity test between the two tools that shape-doctor.py: submodule_config and gitmodules_entry read a holder's .gitmodules through one git config -z --list, so a section name holding a space or a byte that is not UTF-8, or a url holding NEL or U+2028, gives the doctor the mount url git and make siblings use, under any locale (#668) #700's other half adds.
siblings.py is a pinned family-root copy. Changing it costs each holder an upstream-changed row, and the PR body says so.

Lands after: #700.

CLAIMED — lane openRepoShape-3, session b30e8b49-f249-4668-adb1-1dbfa93c09e1@Eagle, 2026-10-16T17:29:50Z, for #733

Logged in lanes/log/openRepoShape-3.md at fdc25977739ff191f66048c5c5f3388f9ae6eddb.

The three reads (lane-collision-protocol Rule 1):

1. existing `CLAIMED —` comments on opensoft/openRepoShape#733:
   none
2. `gh pr list --repo opensoft/openRepoShape --state all --search "#733"`, kept where the row or its body names `#733`:
   none
3. `git ls-remote --heads git@github.com:opensoft/openRepoShape` branches naming `733` as a whole token:
   none

(lane openRepoShape-3, 2026-10-16T18:05:42Z) Three more cases for #733, from RV-c's re-verification of #700 at fca40dfc (issuecomment-6221703634). Each was measured against git submodule init -- <path>.

An unmerged .gitmodules. git's repo_read_gitmodules reads no copy while .gitmodules is unmerged (is_gitmodules_unmerged), so git submodule init exits 128 with No url found.
- The doctor gives the working-tree copy's url (('p', '/r/resolved.git')). Once that file is deleted, it gives HEAD's (('p', '/r/main.git')).
- The doctor's merge row (shape-doctor.py answers COMPLIANT, exit 0, for a root in the middle of a merge (UU contracts/spec-pin.yaml, UU spec, a MERGE_HEAD, git commit refusing), because no row asks git for unmerged paths and recorded_gitlink answers an unmerged gitlink from HEAD, so a root an agent cannot commit is reported compliant, including after the git checkout --ours that a conflicted pin's NEXT names #424) already reports the unmerged path first, so this is not urgent.
- Fix: if git ls-files -u -- .gitmodules is non-empty, gitmodules_entry returns (None, None) before choosing a copy. Add the case to INDEX_COPIES.
The last section giving a path has no url. git says No url found. gitmodules_entry gives ('new', None), so the member is url-less, but mounted_from gives the earlier section's /r/old.git, as siblings.py does.
- mounted_from names the remote make siblings clones from, so agreeing with siblings is its contract. Any change moves both tools together.
- validate-family.py's family.py: add never reads .gitmodules, so a section under ANOTHER name whose path is already members/<P> (a "ghost" section) does not stop it: it exits 0 and commits two sections for one path, validate-family.py says family ok, and a recursive clone takes the LATER section, so with the ghost second it dies fetching the ghost's URL #300 check already refuses two sections for one path.
Reopened sections.
- git orders by the last path key, mounted_from by a name's first key of any kind, and siblings by its first url key.
- Order (a): A has the path; B has the path and a url; A is reopened with a url. Git gives /r/b.git, shape-doctor.py: submodule_config and gitmodules_entry read a holder's .gitmodules through one git config -z --list, so a section name holding a space or a byte that is not UTF-8, or a url holding NEL or U+2028, gives the doctor the mount url git and make siblings use, under any locale (#668) #700 gives /r/b.git, siblings gives /r/a.git.
- Order (b): B has a url only; A has the path and a url; B is reopened with the path. Git gives /r/b.git, while shape-doctor.py: submodule_config and gitmodules_entry read a holder's .gitmodules through one git config -z --list, so a section name holding a space or a byte that is not UTF-8, or a url holding NEL or U+2028, gives the doctor the mount url git and make siblings use, under any locale (#668) #700 and siblings give /r/a.git.
- shape-doctor.py: submodule_config and gitmodules_entry read a holder's .gitmodules through one git config -z --list, so a section name holding a space or a byte that is not UTF-8, or a url holding NEL or U+2028, gives the doctor the mount url git and make siblings use, under any locale (#668) #700's mounted_from docstring (shape-doctor.py:7120-7123 at fca40dfc) holds for plain sections only. Soften it in this issue's PR, and add a reopened section to the parity list.

Lane: openreposhape-3 (openRepoShape-3)

shape-doctor.py: `submodule_config` and `gitmodules_entry` read a hol…

shape-doctor.py and siblings.py: an index or HEAD copy of `.gitmodule…

shape-doctor.py and siblings.py: a `shallow` or `fetchRecurseSubmodul…

tests/test_siblings_inputs.py: `[submodule "a=b"]` mounts and a CR in…

siblings.py: the printed move runs as typed past a taken staging name…

Merge main 40e678e (#805, #690) into l3-733-ignored-sections

siblings.py: a `.gitmodules` section NAME keeps git's bytes, so two n…

validate.yml (assembly-root and family-root templates): the three url…

the family root's bootstrap.py and validate-family.py: the credential…

adopt-project.py (#910, fix round 2): each throwaway `git init` names…

Navigation Menu

Repository navigation

shape-doctor.py and siblings.py: a `.gitmodules` section git ignores (`[submodule ""]`, a name with a `..` component, a url beginning with `-`) is still read as a member, where git says `No url found` and exits 128 #733

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

shape-doctor.py and siblings.py: a .gitmodules section git ignores ([submodule ""], a name with a .. component, a url beginning with -) is still read as a member, where git says No url found and exits 128 #733

Description

Activity

brettheap commented on Oct 16, 2026

brettheap commented on Oct 16, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

shape-doctor.py and siblings.py: a `.gitmodules` section git ignores (`[submodule ""]`, a name with a `..` component, a url beginning with `-`) is still read as a member, where git says `No url found` and exits 128 #733