scripts/check-drift.sh section 2 is the check that a cited path exists, and it is anchored on (src|scripts|tests|tools|docs|experiments|examples|inventories)/ (the anchor= assignment). A bare filename therefore never matches, so a comment that cites a document by name while naming the wrong one is invisible to it.
src/Makefile:5 does exactly that:
# new one does not. The measured builds in REPORT.md used debian:bullseye-slim
There is no REPORT.md. The measured record is docs/report/README.md. The line also names the floating debian:bullseye-slim, which is the tag scripts/build.sh and scripts/suite-lib.sh deliberately stopped using. The check's own comment says it caught docs/REPORT.md in this file once; the path lost its docs/ prefix and dropped back under the anchor.
Fix: point the comment at docs/report/README.md and the dated tag, and widen the cited-path check to catch a bare *.md filename in a comment (or at least the names of the documents the tree actually has), so this class stops recurring.
scripts/check-drift.shsection 2 is the check that a cited path exists, and it is anchored on(src|scripts|tests|tools|docs|experiments|examples|inventories)/(theanchor=assignment). A bare filename therefore never matches, so a comment that cites a document by name while naming the wrong one is invisible to it.src/Makefile:5does exactly that:There is no
REPORT.md. The measured record isdocs/report/README.md. The line also names the floatingdebian:bullseye-slim, which is the tagscripts/build.shandscripts/suite-lib.shdeliberately stopped using. The check's own comment says it caughtdocs/REPORT.mdin this file once; the path lost itsdocs/prefix and dropped back under the anchor.Fix: point the comment at
docs/report/README.mdand the dated tag, and widen the cited-path check to catch a bare*.mdfilename in a comment (or at least the names of the documents the tree actually has), so this class stops recurring.