Skip to content

Commit c6f6397

Browse files
author
igor
committed
Remove misclassified cncmill1-context entries from linuxcnc/memory/
Operator caught two leaks that my Phase 4c grep missed: both files were classified as 'linuxcnc fork knowledge' in Phase 1, but their content is heavily cncmill1 work, not anything about the linuxcnc fork itself. The straddle rule should have applied — cncmill1 work that uses the fork stays in lab/cncmill1/memory/, not linuxcnc/memory/. Removed: - memory/reference_runtime_latest.md Leaked content: 'auto-discovers linuxcnc repo as sibling of lab repo root', src/linuxcnc/screens/qtdragon_hd/{release,latest}/ paths (cncmill1/src structure, not anything in the linuxcnc fork). The RUNTIME=latest mechanism is implemented in cncmill1's Makefile, not in this fork. - memory/reference_test_timings.md Leaked content: spindle-slip-fix reference, base/logic/ui/panels test suite (cncmill1's pyramid, not the linuxcnc fork's runtests), panel phases (spindle/tuning/zcal/lube — cncmill1 subsystems), panel button labels (Move-to-Safety, Apply, Test Gains — cncmill1 panel widgets), test_helpers.sh check_atspi_health (cncmill1's test framework). None of this describes the linuxcnc fork itself. Both files are being added to lab/cncmill1/memory/ in a follow-up commit on the lab repo, where they belong per the straddle rule. The deletions are an overlay commit on the public fork — both files were briefly visible on the session branch since Phase 4c (commit 8b46870), and the commit history preserves them. The public fork policy is no force-push, so the past visibility cannot be undone, only overlaid. The leaked content is not catastrophic (no credentials, no PII) but is private project context that shouldn't have been pushed. Lesson: the public-fork grep pattern needs to catch contextual project terms (spindle-slip, M3/M4, panel phase names, run-dev-sim, INI parameter names, test_helpers.sh) in addition to literal path strings (cncmill1, lab/, ~/is-dev). Grep alone isn't enough — needs human review of every file going to the public fork, even files that pass the path-string grep.
1 parent 8b46870 commit c6f6397

3 files changed

Lines changed: 0 additions & 88 deletions

File tree

MEMORY.md

Lines changed: 0 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -2,8 +2,6 @@
22

33
- [Fork one-time setup](memory/reference_fork_setup.md) — 8-step setup procedure for a fresh clone (Debian, debian/control, build/runtime deps, configure, agent-helpers hooks, setuid, RT tuning)
44
- [halcompile workflow](memory/reference_halcompile_workflow.md) — compile-and-symlink pattern for sudo-once unattended `.comp` recompiles, halcompile naming gotcha, `variable` keyword usage
5-
- [RUNTIME=latest workflow](memory/reference_runtime_latest.md) — default is the locally-built fork; `release` keyword switches to system-installed; rip-environment auto-source semantics
6-
- [Test timings](memory/reference_test_timings.md) — healthy sim suite baselines; if panel tests exceed 150s suspect AT-SPI hang
75
- [Upstream PR workflow](memory/reference_upstream_pr_workflow.md) — full procedure for `agent-helpers/upstream-pr-prep.sh`, prerequisites, worktree rationale, build-flag CI parity, worked example PR #3933
86
- [Teleop velocity leak](memory/reference_teleop_velocity_bug.md) — upstream `axis_sync_teleop_tp_to_carte_pos` doesn't reset `curr_vel`/`curr_acc`; post-homing drift; fork master commit `f40e8e64bc`, upstream PR #3933
97
- [Public fork hygiene](memory/feedback_public_fork.md) — never reference private repo names, paths, or links from any committed file in this fork

memory/reference_runtime_latest.md

Lines changed: 0 additions & 19 deletions
This file was deleted.

memory/reference_test_timings.md

Lines changed: 0 additions & 67 deletions
This file was deleted.

0 commit comments

Comments
 (0)