edfcore 0.2.49 → 0.2.51

package/CHANGELOG.md

@@ -6,6 +6,37 @@ 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.
 
+## 0.2.51
+
+- **Added** the tests only a real, large recording can support, against the 22-hour sleep-edfx
+  polysomnogram. Three claims edfcore advertises are invisible on a hundred-byte fixture:
+  - **Envelope decimation is faithful.** 7,950,000 samples reduced to 1,000 buckets, compared with
+    an EXHAUSTIVE reduction of every one of them — not a spot check. This is the case the feature
+    exists for, and until now it was only demonstrated on 40 records.
+  - **The bucket grid does not move with the read chunk size.** 265 chunks versus a handful. A fold
+    that computed its bucket on the chunk's grid rather than the run's would diverge here and agree
+    on a two-record fixture.
+  - **Memory is bounded by the chunk, not the window.** The 22-hour envelope is produced under a
+    512 KiB budget; materialising the window would need ~32 MB, so the budget makes the claim
+    falsifiable rather than asserted.
+- Also pins that a 100 Hz and a 1 Hz channel keep their own sample grids across the whole file,
+  checked at the LAST 30 seconds — where a shared-grid error is largest, and nowhere else.
+- **Fixed** an unnecessary biome suppression added in 0.2.50 that was itself the only warning in
+  the tree.
+
+## 0.2.50
+
+- **Added** a corpus-coverage report that always runs. Every other test in `tests/corpus/` skips
+  when the files are absent, which is right — a fresh clone must stay green and offline, and none
+  of the corpus is redistributed. But a skipped test is indistinguishable from a passing one in a
+  summary line, and the corpus is where this project's strongest claims live: `1487 passed` reads
+  the same whether the bit-for-bit check against a 22-hour clinical recording ran or not. It now
+  says which state the run is in, once, in a line a reader sees.
+- It also checks the parts that need no corpus: every manifest entry names a source, a licence and
+  a SHA-256, no committed golden refers to a file the manifest no longer lists, and every fetched
+  file has a parity golden. A golden that has drifted from the corpus definition still looks like
+  coverage, which is the failure this catches.
+
 ## 0.2.49
 
 - **Added** annotation, start-date and geometry parity on the real corpus, alongside the sample

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 = "0.2.49";
+export declare const VERSION = "0.2.51";
 //# 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 = '0.2.49';
+export const VERSION = '0.2.51';
 //# sourceMappingURL=constants.js.map

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "edfcore",
-  "version": "0.2.49",
+  "version": "0.2.51",
   "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 = '0.2.49';
+export const VERSION = '0.2.51';