Repository navigation
breaking change: tool is the type, published-verdict is an overlay - #2
Merged
Merged
Conversation
added 2 commits
August 29, 2026 17:21
"Under ~60 characters" rested on one 58-character line reported as clipped, with no record of the width it was seen at. Rendering a ruler of increasing line lengths and screenshotting it says the desktop column fits about a hundred characters and a 390px phone about forty-two, so 60 was safe at neither end: it clips on the phone and constrains nothing on the desktop. The original observation does not reproduce at any width our readers use. The rule the incident actually earned survives — a fence clips rather than wraps, silently — and it now carries the measurement and the retraction instead of a number nobody took.
Yesterday's third type answered the wrong question. Its own placing table said a research line is not something a human opens as a product — but a developer runs `pipx install`, then invokes it, which is opening it as a product. And the reason it did not fit APPLICATION.md was never that it published verdicts: that file assumes a consumer product, with distribution channels, an update mechanism, a telemetry policy, a first-run experience and localisation, none of which a command-line tool has. The missing category was **tool**, and the verdict discipline was standing in for it. They are two different things, and only one of them is a type. TOOL.md carries what every developer-run tool owes — the CLI surface is the public API, exit codes separate "no" from "did not run", one command proves it offline, it refuses rather than guesses, artefacts carry provenance from a single version record, long runs stay watchable. PUBLISHED-VERDICT.md keeps what is genuinely orthogonal — a repo whose product is a claim that could come out negative — and moves to overlay/, where a library could adopt it too. Overlays are a new layer in §13: types narrow, overlays add. The two §7.1 bullets are no longer "not applicable to a type"; they are suspended by an overlay that names them.
There was a problem hiding this comment.
Pull request overview
This PR updates the Codechu standards docs to (1) retract the previously stated “~60 characters” command-line column limit (keeping the underlying “code fences clip” rule, now with measured widths), and (2) reframe “published verdict” as an overlay applicable on top of a project type while adding a new Tool project type.
Changes:
- Replace the “~60 characters” command-line guidance with a measured explanation of code-fence clipping and real-world column widths (desktop vs phone).
- Introduce
project-type/TOOL.mdand update type listings to include Tool. - Add an
overlay/layer with aPUBLISHED-VERDICToverlay and update STANDARDS to reference overlay-based rule suspension.
Reviewed changes
Copilot reviewed 6 out of 6 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
| STANDARDS.tr.md | Updates Turkish standards to reference the Published-Verdict overlay for §7.1 suspensions and replaces the “~60 chars” rule with measured clipping guidance. |
| STANDARDS.md | Updates English standards for overlay-based suspension, retracts the “~60 chars” rule with measured guidance, and adds “Overlay” as a conventions layer. |
| project-type/TOOL.md | Adds a new language-agnostic Tool project type specification and its README skeleton. |
| project-type/README.md | Updates project type list to include Tool and introduces overlays as additive disciplines. |
| overlay/README.md | Introduces the overlay concept and documents the Published-Verdict overlay entry point. |
| overlay/PUBLISHED-VERDICT.md | Reframes published verdicts as an overlay (not a project type) and updates related guidance/checklists accordingly. |
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
Comment on lines
464
to
466
| This file is the language-agnostic, org-wide layer. Two narrower | ||
| layers extend it: | ||
|
|
Comment on lines
39
to
+42
| Read in order: STANDARDS → project-type → lang. Each layer narrows. | ||
| Disciplines that *add* rather than narrow — a repository whose product | ||
| is a claim, for instance — live in [`overlay/`](../overlay/) and are | ||
| adopted on top of a type. |
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
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 changes, one per commit. Both correct something merged yesterday.
The column number is retracted. §7.1 said command lines belong
"under ~60 characters", on the strength of a single 58-character line
reported as clipped — with no record of the width it was seen at.
Rendering a ruler of increasing line lengths (pandoc +
github-markdown-css, headless Chrome at a fixed window) says the
desktop column fits about a hundred characters and a 390px phone about
forty-two. Sixty was safe at neither end, and the original observation
does not reproduce at any width our readers use. The rule the incident
actually earned survives — a fence clips rather than wraps, silently —
and it now carries the measurement and the retraction instead.
Tool is the type; published-verdict is an overlay. Yesterday's
RESEARCH-LINE.mdanswered the wrong question. Its own placing tablesaid such a repo is not something a human opens as a product, but a
developer runs
pipx installand then invokes it, which is opening itas a product. And the reason it did not fit
APPLICATION.mdwas neverthat it published verdicts: that file assumes a consumer product —
distribution channels, an update mechanism, a telemetry policy, a
first-run experience, localisation — none of which a command-line tool
has. The missing category was tool, and the verdict discipline was
standing in for it.
project-type/TOOL.mdcarries what every developer-run tool owes: theCLI surface is the public API, exit codes separate "no" from "did not
run", one command proves it offline, it refuses rather than guesses,
artefacts carry provenance from a single version record, long runs stay
watchable.
overlay/PUBLISHED-VERDICT.mdkeeps what is genuinelyorthogonal — a repo whose product is a claim that could come out
negative — where a library could adopt it too.
Overlays are a new layer in §13: types narrow, overlays add. The two
§7.1 bullets are no longer "not applicable to a type"; they are
suspended by an overlay that names them.
Ripple: a repo classified as a research line yesterday reclassifies as
a tool that adopts an overlay. TR mirror updated. Old name grepped
repository-wide, no dead cross-references.