edfcore 0.5.70 → 0.5.72

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 = "0.5.70";
+export declare const VERSION = "0.5.72";
 //# 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 = '0.5.70';
+export const VERSION = '0.5.72';
 //# sourceMappingURL=constants.js.map

package/docs/CHANGELOG.md

@@ -6,6 +6,49 @@ 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.
 
+## 0.5.72
+
+- **Added** the case none of the four caller-supplied callbacks documents: what happens when yours
+  throws. edfcore calls a `ByteSource.read`, an `onProgress` on each of the two traversals, and a
+  predicate on the two matching helpers. Three of the four say what they should RETURN and none of
+  the four says what a throw does — which is not exotic: a progress callback writes to a DOM node
+  that has been removed, a label predicate calls `toLowerCase` on a signal with no label.
+- Two answers matter and they pull opposite ways. The error must arrive UNCHANGED — the same
+  object, not wrapped, not turned into a diagnostic — because wrapping it makes the caller's own bug
+  look like a problem with the file, and `isEdfError` has to say false for it. And the recording
+  must SURVIVE it: every one of these calls happens partway through something with state, and a
+  throw unwinds through all of it.
+- `when-your-callback-throws.test.ts` checks both for the three that had nothing;
+  `source-contract.test.ts` already covers the source. After a callback throws mid-scan, the same
+  recording rebuilds to the same index, validates to the same verdict, and leaves the annotation
+  array it was given untouched — each compared against a recording that never saw a failure.
+- One check is about the failure a memoised promise would produce: a second attempt has to reach
+  the callback again rather than a remembered rejection, or one bad progress callback breaks the
+  operation for the life of the recording. Wrapping the progress call in a `try`/`catch` that
+  swallows fails three of the nine.
+
+## 0.5.71
+
+- **Added** the case every diagnostic test avoids: a file with more than one thing wrong with it.
+  Each code is demonstrated on a fixture broken in exactly one way, which is how you show which
+  defect produces which code — and real files are not like that. Defects come in families, because
+  they come from one mistaken program.
+- The header parse is a single pass over shared state: one sink, a pinned check order, and checks
+  that read fields other checks have already judged. That is the shape in which one defect swallows
+  another — an early return on the first bad field, a check skipped because its input was already
+  reported unusable — and the file would still open with a report that looked complete.
+- `a-second-defect-hides-nothing.test.ts` applies seven independent corruptions alone, then in all
+  twenty-one pairs, then in both orders, then all seven at once. Making the recording-identification
+  check skip itself once anything else has been reported fails six of them.
+- Six of the seven are independent in the file. The seventh is not, and finding that out is half of
+  what this is worth: a reserved field that is not a recognised marker makes the file plain EDF, and
+  `parse.ts` passes `edfPlus: variant.isPlus` into both identification grammars — so on that file
+  they are not checked and their two codes correctly disappear. Not masking; the file no longer
+  deserves them. Worth pinning anyway, because the visible effect is that one wrong five-byte field
+  silently stops the patient and recording identification from being examined at all, and nothing
+  in the report says so. Fixing the reserved field alone brings both codes back, which is what makes
+  it the cause rather than an ordering coincidence.
+
 ## 0.5.70
 
 - **Added** the third thing a caller hands edfcore that edfcore does not own. The buffer is checked

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "edfcore",
-  "version": "0.5.70",
+  "version": "0.5.72",
   "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 = '0.5.70';
+export const VERSION = '0.5.72';