edfcore 0.6.15 → 0.6.16

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 = "0.6.15";
+export declare const VERSION = "0.6.16";
 //# 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 = '0.6.15';
+export const VERSION = '0.6.16';
 //# sourceMappingURL=constants.js.map

package/docs/CHANGELOG.md

@@ -6,6 +6,30 @@ 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.
 
+## 0.6.16
+
+- **Added** the check that the five source constructors are interchangeable, which is what
+  `data-sources.md` promises and what nothing executed. Each was tested for the thing it does —
+  `file-source.test.ts` for descriptors and stat, the `http-*` files for Range and 200s and
+  concurrency, `cache-invisible.test.ts` for the block cache — and nothing ran the reading API over
+  all of them and compared the answers.
+- That is the claim a caller depends on: develop against `byteSource` in a test, ship `httpSource`
+  in a browser and `fileSource` in a batch job, and expect one recording. The seams are real.
+  `fileSource` reads through a descriptor at an offset, `httpSource` splits a read into Range
+  requests, `cachedSource` serves from blocks that never line up with a record — three paths to the
+  same bytes, each a place an off-by-one would show through one source and no other.
+- Header, timeline, segments, gaps, annotations, records, window, envelope and the validation
+  report, over every shape in the matrix, through all five. They agree.
+- `bytesRead` is compared exactly, cache included. A block cache reads whole blocks, but a full
+  conformance sweep reads the whole file whatever the blocking and no source may read it twice —
+  checked with a block size small enough to split every record. The over-reading a partial window
+  provokes is `cache.test.ts`'s subject and is not this one.
+- Every source is closed, in a `finally`. `fileSource` holds a descriptor and `close()` is optional
+  on the interface, so a sweep that opens one per shape per spelling leaks them — which on Node 26
+  is not a warning but an uncaught `ERR_INVALID_STATE` when the handle is collected. The three-way
+  Node matrix in CI is what said so, on a run that passed on 22.12 and 24.
+- No behaviour changed. Nothing disagreed.
+
 ## 0.6.15
 
 - **Fixed** four copies of the definition of EDF field padding. `trimEdfField` states the rule and

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "edfcore",
-  "version": "0.6.15",
+  "version": "0.6.16",
   "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 = '0.6.15';
+export const VERSION = '0.6.16';