2026-09-03 - (/loop) The encoder measurements say the opposite, and thirty names go back

declared          947  ->  917
names-only        273  ->  243 across 29 libraries

Asked to check whether the thirty-three encoder symbol measurements are claimable now that libSceVencCore and libSceVideoRecording are declared. No - and the premise was inverted, mine included.

unresolved = 0x0 means it did not resolve

if (addr != NULL) {
    obs_report_measure("106-encoder/symbols", name, "vaddr",  …);
    obs_report_measure("106-encoder/symbols", name, "handle", …);
} else {
    obs_report_measure("106-encoder/symbols", name, "unresolved", 0, "status");
}

The record is emitted in the null branch, after the lookup has tried the kernel export table, every loaded handle and 0x2001. The 0 is filler for a status field. I had read it as "zero unresolved" and told the user every one of those symbols resolves on the console.

Two independent confirmations: there is not one vaddr or handle measurement in the whole encoder group - every probed symbol failed - and obSCEne's related-libs check records six of those libraries as absent. The reasons for the libraries in this file were already correct; only the reasons for the symbols inside them were backwards.

So thirty declared names had no provenance and are withdrawn

D505 set the standard in its own text - every name "read out of a real module's import table" or measured resolving. Against it: no guest in the corpus imports or calls libSceVencCore, and nothing resolved, so its 24 names rested on obSCEne's candidate list alone. Six of libSceVideoRecording's ten were the same. Withdrawn. The four the corpus shows modules calling stay.

That is a bad declaration committed inside the decision that exists to prevent bad declarations. A name from a candidate list puts a NID in the table no import will ever match and inflates a surface figure with functions nothing has been seen to want.

The thirty-three reasons are rewritten

The old text - "whether a symbol resolves inside a video-encoder library" - was not vague, it was backwards, and it would have justified a claim in the wrong direction. They now say what they record: a non-resolution under one capture's application category, the same shape as the sysmodule refusals (D503), unblocked by the category-0 capture already in obSCEne's backlog 022.

The cheap check to add

Rule 6 is read the probe before asserting against its numbers. What it needs is the direction test: a field named unresolved/absent/refused is emitted by a failure branch, so its presence is the finding and its value is filler. Reading 0x0 as a count rather than as a marker is D497's mistake one field name further out.

State

cargo test --workspace green - 119 suites, 1994 tests, 0 failures. clippy --tests clean, fmt clean, identity scan clean on both.

Nothing committed. The day holds worklogs 292-355 and D466-D506.

Next: the differential's uncovered functions (strtok_r, strtof, sprintf, vsnprintf), and the two singletons still untried - 120-measure/timer-ratio:tsc_hz_calibrated and 035-libc/fpu-environment.