Repository navigation
Conversation
CheckCMDBDataForUnexpandedVars() rejected a whole 'vars'/'variables'/ 'classes' section on the first entry containing an unresolved variable reference, discarding every other entry in that section along with it. Checked per entry instead, inside each section's own install loop, matching this file's existing skip-and-continue idiom for other per-entry problems (an invalid variable specification, a missing 'value' field). New acceptance test: one good and one bad entry in the same 'vars' section, asserting the bad entry is still rejected and named, and the good entry is installed anyway. Discrimination proved: fails on the unpatched function, passes with the fix; full 00_basics/06 suite still 14/14. Changelog: Title Ticket: CFE-4720
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
|
Linux gate backfill (Ubuntu 24.04 aarch64 container, root,
No failures, no soft failures. |
|
Closing as superseded: retired under djbclark's ruling of 2026-11-14 (B-5b "Replace"). The fork's CMDB loader is now all-or-nothing: a file with any invalid entry installs nothing (R13, 2026-11-02). That landed in the cmdb-fail-closed series, finished in #98, and is merged into tendcf-integration at 6505f48. Skipping only the bad entry goes the opposite way, so this change will not be offered upstream. The branch is kept, not deleted. |
Skipped only the bad entry when CMDB data contains a variable reference
One entry holding an unresolved variable reference rejected the whole
vars,variablesorclassessection. Now only that entry is rejected (and named); the others are installed, as for the other per-entry problems in this file. This relaxes the fail-closed behaviour of the existing check, so it needs a maintainer's judgment. Stacked on #57.Fixes #9
Tests: new acceptance test 16-variable-references-good-entry-survives pass on macOS (
06_host_specific_data: 39 pass, 0 fail, 1 soft fail); Linux--enable-debuggate on this head: unit check 78/78,29_simulate_mode7/7 (the new acceptance test ran on macOS only).Adversary review (omp Claude Opus, read-only): APPROVE. P2 (design): a host can now end up with a mixed CMDB state that its author did not intend; the only existing upstream test of this (
10-variable-references.cf) cannot detect the change. P3: the section-levelFailed to load CMDB dataline no longer appears when entries are skipped.AI-assisted; reviewed by djbclark only at the level of this fork PR.
🤖 Generated with Claude Code