docs(skill): 429-430 -- a broken corpus is not a control, and a wobble count is not a population - #3008
Merged
Conversation
…e count is not a population 429. Repairing #2997 I also renamed a `_` capture, copying the C emitter, and produced `register `__t27_i' unknown` in every file it touched. `iverilog` read 380 before and 380 after with an identical accepted set, because all 36 carriers were already inside the failing 201 -- on the very defect being repaired. No acceptance number could move whatever I emitted into them. That is the inverse of "acceptance cannot see a defect that compiles", and worse: the corpus was not a weak control but no control, and the aggregate was reassuringly flat. A probe that RUNS caught it in a second. 430. The first determinism harness read 1 / 3 / 2 differing files; the second read 4 / 2 / 4 on the same tree with the same binary. Neither is wrong -- which files wobble is a draw, so a count taken once measures the draw. The union of NAMES is stable and small: four specs, two of them in all three backends. Three counts read as three independent emitter defects; four shared names read as one shared path -- a different and tractable investigation. `tri skill check`: 394 sections, no number used twice. Refs #2997, #3006
Contributor
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Two sections, both earned this pass.
429 — a corpus of already-broken files is not a control. Repairing #2997 I also renamed a
_capture, copying the C emitter. C declares its counter in theforheader; the Verilog backend hoists the declaration from a different function, so the rename producedregister \__t27_i' unknownin every file it touched — andiverilog` read 380 before and 380 after with an identical accepted set, because all 36 carriers were already inside the failing 201, on the very defect being repaired. This is the inverse of "acceptance cannot see a defect that compiles" and worse: the corpus was not a weak control but no control at all. A probe that RUNS caught it in under a second.430 — a count of a wobble measures the draw. The harness that opened #3006 read 1 / 3 / 2 differing files; a second harness read 4 / 2 / 4 on the same tree with the same binary. Neither is wrong. Only the union of names is stable — four specs, two wobbling in all three backends. Three counts read as three independent emitter defects; four largely-shared names read as one shared path four specs reach.
tri skill check: 394 sections, no number used twice.Refs #2997, #3006
🤖 Generated with Claude Code