edfcore 0.5.41 → 0.5.43

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 = "0.5.41";
+export declare const VERSION = "0.5.43";
 //# 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 = '0.5.41';
+export const VERSION = '0.5.43';
 //# 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.
 
+## 0.5.43
+
+- **Added** the two halves of what a sample width decides, written side by side. EDF stores each
+  sample in two bytes and BDF in three; that is the whole difference between the families as far as
+  the data records go, and `decodeDigital` keeps it where it belongs. The suite tested each width
+  against expected values and never one against the other.
+- For a value both widths can hold, the two files disagree about everything except the number: the
+  record is half again as long, the declared digital range is 256 times wider, `bitValue` differs by
+  the same factor — and the decoded samples are bit-identical. That is "the width is an encoding",
+  asserted where it could fail.
+- For a value only BDF can hold, the EDF file does not hold it. Two bytes cannot carry −300,000, so
+  what is written is the low sixteen bits and what comes back is those bits sign-extended: 27,680.
+  edfcore reports it with no diagnostic and no `outOfDigitalRangeCount`, which is correct and worth
+  writing down — the file is well formed, 27,680 is inside its declared range, and the loss happened
+  before edfcore saw a byte. Nothing in an EDF header can record that a writer had a number it could
+  not store.
+- The whole low-bits sequence is asserted rather than the first value, and the same waveform in the
+  two families is shown to be two different recordings carrying identical diagnostics — which is the
+  point. `out-of-range.test.ts` covers the other case, a sample outside the DECLARED range in a
+  width that can hold it; this is the case where the width itself is the limit.
+
+## 0.5.42
+
+- **Added** the transformation that changes what the samples MEAN without changing the samples.
+  Rewriting `physicalMinimum`/`physicalMaximum` in the header leaves every stored integer identical
+  — `decodeDigital` never looks at the range and `toPhysical` never looks at the bytes — and moves
+  everything derived from it onto the new scale. Nothing tested the two against each other:
+  `physical-values.md` is checked against printed numbers and `out-of-range.test.ts` against the
+  digital path, and the relationship between them was prose.
+- The interesting half is the part that does **not** hold. Multiplying the declared range by ten
+  multiplies `bitValue` by exactly ten and leaves `offset` untouched — both exact in float64 — and
+  yet a quarter of the converted values are not exactly ten times their counterpart.
+  `bitValue * (offset + digital)` is a float64 multiply, and scaling an operand is not the same
+  operation as scaling the result.
+- That is worth pinning because it is a mistake a caller makes on purpose. Given microvolts and
+  wanting millivolts, the obvious move is to divide the values, and the obvious check is that
+  dividing the declared range agrees. It does not, in the last place, for a quarter of the range —
+  and the last place is what `physical-values.md` spends a page refusing to be casual about.
+- So four things are asserted together: the digital path is bit-identical under the transformation
+  and so are the structural numbers and the observed digital range; the scale moves by exactly the
+  factor the declaration moved by; every converted value is on its own declared scale exactly; and
+  the rescaling identity fails, by at most one unit in the last place and for neither none nor all
+  of the values. Distributing the multiply in `toPhysical` — an algebraically equivalent
+  rearrangement — fails three of them.
+
 ## 0.5.41
 
 - **Added** a metamorphic test: reordering the signals in a header changes where the bytes are and

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "edfcore",
-  "version": "0.5.41",
+  "version": "0.5.43",
   "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 = '0.5.41';
+export const VERSION = '0.5.43';