edfcore 0.5.39 → 0.5.41

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 = "0.5.39";
+export declare const VERSION = "0.5.41";
 //# 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 = '0.5.39';
+export const VERSION = '0.5.41';
 //# 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.
 
+## 0.5.41
+
+- **Added** a metamorphic test: reordering the signals in a header changes where the bytes are and
+  nothing else. EDF interleaves — every data record holds each signal's samples end to end in header
+  order, and a signal's block begins at the sum of the widths declared before it — so a channel's
+  position decides the offset every read of it is computed from, and decides nothing about what that
+  channel contains.
+- That makes signal order a transformation with unusual reach: it changes every offset in the
+  de-interleaving arithmetic and must change no value anywhere. A fixture with one signal, or with
+  several of equal width, cannot tell the difference. The arithmetic only has room to be wrong when
+  the widths differ — which is exactly the file the format exists for, EEG at 256 Hz beside a
+  temperature channel at 1 Hz.
+- The suite tested reading against expected values, per fixture; nothing tested it against itself
+  under a transformation. A `recordByteOffset` computed from the wrong running total, an off-by-one
+  in the signal loop, or a decode that assumed uniform width would survive every fixture in the
+  suite and fail here.
+- Both directions are asserted, because half of it would be worthless. The offsets really do move —
+  three orders give three different offset sets, and `EEG Fpz-Cz` moves from byte 0 to byte 18 while
+  `Temp rectal` moves from 80 to 0 — and every per-label result stays identical: samples, physical
+  values, envelope buckets, sample counts, the range the sweep observes, and what `getSignal` and
+  `findSignals` return.
+- The last block is the property over arbitrary widths and arbitrary rotations, with a constant seed.
+
+## 0.5.40
+
+- **Added** the resolution of every index and byte offset edfcore publishes. The API is full of
+  numbers that are addresses rather than measurements: `signalIndex` names a row of
+  `header.signals`, `recordIndex` a record, `firstSampleIndex` a position on a signal's own grid,
+  and `byteOffset`/`byteLength` a range a reader is expected to take to a hex editor. A measurement
+  that is wrong is wrong; an address that is wrong sends someone to look at the wrong bytes.
+- They were checked one at a time where they are produced and nowhere as a class. Every object every
+  entry point returns is now walked over the eight `AWKWARD` shapes and a file with a gap — 1,500
+  numbers across 51 field names — and each is resolved against the file it came from. A
+  `signalIndex` must name a signal whose own `index` is that number; a `recordIndex` a record the
+  file has; `firstSampleIndex + sampleCount` must land inside the signal's `sampleCount`.
+- The strongest check re-decodes. A chunk's `byteOffset` and `byteLength` are used to slice the
+  fixture and decode it again, and the samples that come back must be the samples the chunk carries.
+  A range check would pass on an offset that is inside the file and points at the wrong record.
+- **Found while writing it:** that re-decode was vacuous on every shared fixture. `writer.ts`
+  defaults its sample generator to `(_record, index) => index % 100` — the same ramp in every record
+  — which is right for a fixture about a header and makes a byte-offset check unable to fail: a
+  chunk reporting the offset of the wrong record decodes to identical samples. The resolution now
+  runs against a file whose every record is distinguishable, asserts that it is before relying on
+  it, and shows that shifting the offset by one record changes what comes back.
+
 ## 0.5.39
 
 - **Added** the rule that every array edfcore hands back is frozen. `src/` calls `Object.freeze`

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "edfcore",
-  "version": "0.5.39",
+  "version": "0.5.41",
   "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 = '0.5.39';
+export const VERSION = '0.5.41';