edfcore 0.4.493 → 0.4.495

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 = "0.4.493";
+export declare const VERSION = "0.4.495";
 //# 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 = '0.4.493';
+export const VERSION = '0.4.495';
 //# sourceMappingURL=constants.js.map

package/docs/CHANGELOG.md

@@ -6,6 +6,39 @@ 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.
 
+## 0.4.495
+
+- **Added** a test that measures the two cost tables on `large-files.md`. That page is the
+  random-access claim made concrete — it names two files by their exact geometry and prints, byte
+  for byte, what opening one and reading ten seconds out of the other costs. Four read ranges with
+  their offsets, two totals, a percentage, a sample count, an overread factor, and the highest byte
+  either read touches. None of it had ever been run.
+- `documented-read-counts.test.ts` checks the read COUNTS `api-reading.md` states in prose. This
+  page states offsets and lengths, which is the stronger claim and the one a reader budgets
+  against: an HTTP round trip is priced by where a range starts and how long it is, not by how many
+  there are.
+- Neither file is built. One is 29 MB and the other 442 MB, and both claims are about which bytes
+  are ASKED for, so a source that reports the length, serves a header of the right geometry and
+  records the ranges answers it exactly.
+- Every figure comes out of the page, so the two cannot drift apart in either direction. All of
+  them hold: the four open reads land on the offsets printed, the ten-second window is one
+  153,600-byte range at 221,191,936, and the far end of the recording is never addressed.
+
+## 0.4.494
+
+- **Added** a test that produces the refusal `large-files.md` prints. That page's budget section is
+  a transcript — three field values and the whole error message, wrapped over three lines, for a
+  named file — and nothing had compared any of it with what `readRecords` throws.
+- A message reworded in `io/read.ts` leaves a transcript on the site that no version of edfcore has
+  ever produced, and the transcript is the part a reader searches for when they hit the error.
+  `budget-boundary.test.ts` owns the rule; this owns the page.
+- The file on that page is 442 MB and is never built. `readRecords` refuses before it allocates or
+  reads, which is the property the section is about, so a source reporting the length and serving
+  the header is all this needs — and the last case counts the bytes to prove the refusal really did
+  come first, rather than after a 442,368,000-byte read of zeros.
+- Every number is read out of the page, including the geometry in the sentence that introduces the
+  file, so a fixture drifting from the prose fails rather than agreeing about the wrong recording.
+
 ## 0.4.493
 
 - **Added** the two edges of `readHeader`'s prefetch decision, both of which could be moved with

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "edfcore",
-  "version": "0.4.493",
+  "version": "0.4.495",
   "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 = '0.4.493';
+export const VERSION = '0.4.495';