edfcore 0.5.49 → 0.5.51

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 = "0.5.49";
+export declare const VERSION = "0.5.51";
 //# 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 = '0.5.49';
+export const VERSION = '0.5.51';
 //# sourceMappingURL=constants.js.map

package/docs/CHANGELOG.md

@@ -6,6 +6,51 @@ 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.
 
+## 0.5.51
+
+- **Added** the promise `annotations.md` makes about partial reads, and the thing it costs.
+  `readAnnotations` "supplies the rebasing origin from the timeline, so a partial range answers the
+  same as a whole-file one" — and could not derive that origin on its own, because "a range that
+  does not contain record 0 has to infer it from an observed onset, which only works while the
+  records in between are contiguous". Before 0.2.28 the pairing the page recommends,
+  `readAnnotations(recording, chunk.records)`, reported the same event a quarter of a second later
+  than a whole-file decode did.
+- Three behaviours, one of them a defect that shipped, and none under test: `readAnnotations` was
+  checked on whole files and on single ranges, and nothing partitioned a file and compared.
+- The property runs first. For any partition of the record range into contiguous pieces, the reads
+  concatenate to the whole-file read — every field of every event, both axes, both exact tick counts
+  and `recordOnsetTicks` — over a contiguous file, one with a quarter-second start offset, and an
+  EDF+D file with a twenty-second hole.
+- Then the three-way comparison that shows the origin doing the work, on the post-gap range of that
+  EDF+D file. `readAnnotations` puts the event at 26.5 s on the record axis. `decodeAnnotations`
+  called directly on the same bytes puts it at 26.75 — its inferred origin lands outside `[0, 1)`,
+  so rebasing switches off and the two axes come back equal, exactly as documented. Passing
+  `startOffsetTicks` by hand restores 26.5. And on the same shape without the hole the inference
+  works unaided, which is what makes the failure about EDF+D rather than about partial ranges.
+
+## 0.5.50
+
+- **Added** three ways a header can be written and mean the same thing. Each is documented, each is
+  reached by real writers, and each was tested for the thing it produces rather than for the
+  equivalence it implies.
+- **A record count of `-1`.** `types.ts` says of the field: "Verbatim. `-1` means the writer never
+  closed the file." `parseHeader` recovers the count from the source's length and says so with
+  `RECORD_COUNT_RECOVERED` and `recordCountSource: 'sourceByteLength'`. `parse.test.ts` checks the
+  recovery; nothing checked that the recovered file is the same recording as one that declares its
+  count — which is the only reason recovering beats refusing.
+- **Either family's annotation label.** `annotations.md`: "edfcore accepts either label in either
+  family, because the label names the channel's *role*." `isAnnotationLabel` is tested on the
+  strings; whether a BDF+ file whose channel says `EDF Annotations` reads its events was not. Both
+  crossings now do, with the same events, the same record onsets and the same data indices — and the
+  label reported verbatim, which is the one thing that differs.
+- **NUL padding.** EDF pads with spaces; real writers pad with NUL, which is why `trimEdfField`
+  exists and why five docblocks in `src/` warn that `String.prototype.trim` does not strip it. The
+  function is tested. That the two paddings give the same header, with no diagnostic between them,
+  was not — and the test asserts that `.trim()` on the raw field really would have got it wrong, so
+  the equivalence is not a coincidence of the fixture.
+- Each is asserted as a whole-recording equivalence — every sample, every event, every derived
+  number — with the one legitimately differing field named beside it.
+
 ## 0.5.49
 
 - **Added** the two things about an EDF+ annotations channel that are decisions of the writer rather

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "edfcore",
-  "version": "0.5.49",
+  "version": "0.5.51",
   "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 = '0.5.49';
+export const VERSION = '0.5.51';