edfcore 0.4.240 → 0.4.244

package/dist/constants.d.ts

@@ -109,5 +109,5 @@ export declare const SIGNAL_FIELD_BLOCK_OFFSETS: {
     readonly reserved: 224;
 };
 /** Published package version. Kept in sync with package.json by a test. */
-export declare const VERSION = "0.4.240";
+export declare const VERSION = "0.4.244";
 //# sourceMappingURL=constants.d.ts.map

package/dist/constants.js

@@ -79,5 +79,5 @@ export const SIGNAL_FIELD_BLOCK_OFFSETS = {
     reserved: 224,
 };
 /** Published package version. Kept in sync with package.json by a test. */
-export const VERSION = '0.4.240';
+export const VERSION = '0.4.244';
 //# sourceMappingURL=constants.js.map

package/docs/CHANGELOG.md

@@ -6,6 +6,62 @@ alone does not tell you whether you were affected.
 edfcore is pre-1.0. Patch releases have carried behaviour changes where the old behaviour was a
 defect; those are called out below.
 
+## 0.4.244
+
+- **Added** a wait for CI before the GitHub release is created, which is what stops a green local
+  run from becoming a version that never reaches npm. `npm run check` runs on the machine cutting
+  the tag, and that is not the same question as whether it passes: twice this week a check was
+  green here and red on every runner — one read a file whose tsconfig lives in
+  `website/node_modules`, which CI does not install, and one required the gitignored
+  `tests/scratch/` to exist. Between them 0.4.231–0.4.236 and 0.4.241–0.4.242 were tagged and
+  never published, eight numbers refused by `publish.yml` long after this script had exited 0.
+  The script now polls the check runs for the exact commit it pushed and refuses to open the door
+  to npm if any of them fails. That turns a silent hole into a stop with the tag intact and the
+  version still recoverable by `gh release create` — the same recovery 0.4.226 wrote the message
+  for. It gives up after twenty minutes and says so rather than hanging.
+
+## 0.4.243
+
+- **Fixed** the layout-table check added in 0.4.241, which compared `tests/README.md` against the
+  filesystem and so required `tests/scratch/` to exist. That directory is gitignored — it holds
+  throwaway reproductions, and committing one would pin whatever behaviour was current when it was
+  written — so it is present on a machine that has chased a defect and absent from every fresh
+  clone, which is every CI runner. The check passed locally and failed on all three matrix jobs.
+  The table documents `scratch/` precisely *because* it can appear, so the rule is now: every
+  directory present has a row, and every row names a directory that is present or listed as
+  ignored in `.gitignore` — read from that file rather than named here.
+
+  **0.4.241 and 0.4.242 were never released**, for the same reason and in the same way as
+  0.4.231 through 0.4.236: tagged, and the publish run stopped at the failing check. Everything
+  they carried is in this release. Both incidents share one cause — a check that passes on the
+  machine cutting the tag and cannot pass on a runner — and this release was verified against it
+  directly, with `tests/scratch/` and `website/node_modules` both moved aside first.
+
+## 0.4.242
+
+- **Fixed** the link checker's own hand-written inventory, six releases after it was added to
+  catch exactly that. 0.4.236 listed the site's standalone routes — `/`, `/demo`, `/llms.txt` and
+  the rest — as a literal set, so deleting a route would have left the list vouching for it, and
+  the check meant to find dead links would have been the last thing claiming that one was alive.
+  Nothing about the site is written down there now: the pages come from the collection, the
+  standalone routes from the files under `pages/` and `public/`, and the redirects from
+  `astro.config.mjs`. A route is a path with the framework extension removed and nothing else, so
+  `llms.txt.ts` is `/llms.txt` — only the `.ts` comes off — and one assertion pins each shape the
+  derivation has to get right.
+
+## 0.4.241
+
+- **Added** checks on `tests/README.md`, which described this suite with nothing checking that it
+  still did. It is where the root README sends a contributor, and the fixture policy it states is
+  the only reason six binary files are committed to a repository that otherwise builds every
+  fixture in memory. Two of its claims were inventories — a table with one row per directory under
+  `tests/`, and a count of the files under `corpus/golden/` stated four times across the two
+  READMEs — and neither was derived, so a new directory would join the suite and not the table,
+  and the number justifying those committed binaries could drift the way the site's page count did
+  one release ago. Both are read from the tree now. The counts stay spelled out and are read
+  through a word list: `tests/README.md` is prose someone reads start to finish, unlike the
+  one-line parenthetical 0.4.238 turned into a numeral.
+
 ## 0.4.240
 
 - **Extended** the link check added in 0.4.236 to the links that point back at this project by

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "edfcore",
-  "version": "0.4.240",
+  "version": "0.4.244",
   "description": "Modern, typed, zero-dependency reader for EDF, EDF+, BDF and BDF+ biosignal files. Works in browsers and Node with true random access.",
   "keywords": [
     "edf",

package/src/constants.ts

@@ -93,4 +93,4 @@ export const SIGNAL_FIELD_BLOCK_OFFSETS = {
 } as const;
 
 /** Published package version. Kept in sync with package.json by a test. */
-export const VERSION = '0.4.240';
+export const VERSION = '0.4.244';