edfcore 0.5.20 → 0.5.22

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 = "0.5.20";
+export declare const VERSION = "0.5.22";
 //# 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 = '0.5.20';
+export const VERSION = '0.5.22';
 //# sourceMappingURL=constants.js.map

package/docs/CHANGELOG.md

@@ -6,6 +6,48 @@ 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.
 
+## 0.5.22
+
+- **Added** the scrolling table on `large-files.md`, measured. `large-files-cost.test.ts` runs the
+  two tables above it — what opening costs, and what one ten-second window costs. This is the third,
+  and the one that carries an argument rather than a figure: thirty consecutive ten-second windows
+  over an 8-channel EDF+C cost 30 reads and 1,246,800 bytes from a bare source, and 1 read of
+  1,048,576 bytes through `cachedSource`.
+- The sentence under it is why the table is there. "Removing `cachedSource` changes the number of
+  reads and nothing else" is a debugging instruction — if a bug appears with the cache in place,
+  delete the wrapper; if it survives, the cache was not involved — and it is only as good as the
+  "nothing else". The two runs are now compared sample for sample, all 153,600 of them, with the
+  count of non-zero values asserted too: a fixture decoding to zeros throughout would make an exact
+  match mean nothing.
+- The page also says which block the single read is — "the second 1 MiB block. The first was already
+  resident because the header read at open pulled it in, so the reads that come with opening a file
+  are not wasted." That is checked as an offset, because it is what a reader plans an HTTP range
+  budget around, and because a cache issuing one read of the right size at the wrong place would
+  satisfy every other assertion here.
+- The neighbouring claim goes with it: thirty ten-second windows cost thirty reads, and one
+  300-second window over the same records costs one, for the same 1,246,800 bytes.
+
+## 0.5.21
+
+- **Added** the census `reading-signals.md` takes of the derived sample rate. `samplesPerRecord` is
+  authoritative and `sampleRateHz` is "provided for display"; the page argues that with one file —
+  256 samples in a 3-second record is 85.333… Hz, no float holds it, and the error grows with *t* —
+  and then counts how often the two ways of finding a sample disagree: "1,000 of the first 3,001"
+  integer second boundaries, "always by exactly one sample", with the float answer landing one
+  sample early.
+- Nothing computed it. `sample-grid.test.ts` covers the grid functions and `trim-window.test.ts` the
+  window arithmetic, both against exact expectations; neither compares either against
+  `Math.round(t * sampleRateHz)`, which is the expression a reader would otherwise have written and
+  the one the whole section exists to talk them out of.
+- The three printed values are checked first — `85.33333333333333`, `8534` from `trimToWindow`, and
+  `8533` from the float, whose sample really does start at 99.996 s — and then the census is run over
+  all 3,001 boundaries through `trimToWindow` rather than through a reimplementation of it. Each
+  clause of the Note is a separate assertion, because they fail separately: a count that is right
+  with the direction wrong would be a worse page than one with no number at all.
+- The page shows `trimToWindow` twice, and the earlier example is on a 256 Hz file with its own
+  `exact.firstSampleIndex`. Every lookup is scoped to the fence that mentions the derived rate, so
+  the test cannot pass by reading the wrong example.
+
 ## 0.5.20
 
 - **Added** the lifetime contract `api-sources.md` states in five words: of `close?()`, "edfcore

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "edfcore",
-  "version": "0.5.20",
+  "version": "0.5.22",
   "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 = '0.5.20';
+export const VERSION = '0.5.22';