D203
- 105-record, not 170-record: the section
numbers carry the layering and the report contract enforces it
The section was first numbered 170 and registered between
110-modules and 120-measure, because the
argument for its position is a dependency one - it drives the encoder
behind the same output 080-video acquires, so a reader
wants to know whether that output works first.
make check rejected the report: sections are out of
order. The registry is an explicit list and will hold any order
somebody writes; the contract requires the identifiers to ascend, and
the numbers are what carry the layering rather than the list
position.
Renumbering to 105 satisfies both - it sits in the presentation group
after 100-input, which is where the dependency argument
already said it belonged. The two constraints were never in tension;
only the number was wrong.
Renaming check identifiers is normally forbidden (principle 3: they are the key when diffing runs). These had never run anywhere, so there was no history to break - and doing it now is what stops there being any later.