edfcore 0.6.19 → 0.6.21
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/constants.d.ts +1 -1
- package/dist/constants.js +1 -1
- package/docs/CHANGELOG.md +38 -0
- package/package.json +1 -1
- package/src/constants.ts +1 -1
package/dist/constants.d.ts
CHANGED
|
@@ -109,5 +109,5 @@ export declare const SIGNAL_FIELD_BLOCK_OFFSETS: {
|
|
|
109
109
|
readonly reserved: 224;
|
|
110
110
|
};
|
|
111
111
|
/** Published package version. Kept in sync with package.json by a test. */
|
|
112
|
-
export declare const VERSION = "0.6.
|
|
112
|
+
export declare const VERSION = "0.6.21";
|
|
113
113
|
//# sourceMappingURL=constants.d.ts.map
|
package/dist/constants.js
CHANGED
package/docs/CHANGELOG.md
CHANGED
|
@@ -6,6 +6,44 @@ alone does not tell you whether you were affected.
|
|
|
6
6
|
edfcore is pre-1.0. Patch releases have carried behaviour changes where the old behaviour was a
|
|
7
7
|
defect; those are called out below.
|
|
8
8
|
|
|
9
|
+
## 0.6.21
|
|
10
|
+
|
|
11
|
+
- **Added** the check that calling twice gives the same answer. Every function in the reading API is
|
|
12
|
+
documented as a question about a file, and a question about a file has one answer — and nothing
|
|
13
|
+
checked it. The suite calls each of them once per test, which is exactly the shape that cannot see
|
|
14
|
+
state left behind by the first call: a memo filled with the wrong key, a cursor advanced and not
|
|
15
|
+
reset, an array handed out and then written into by the next caller.
|
|
16
|
+
- Two axes, because they catch different things. The same recording asked twice catches state on the
|
|
17
|
+
recording; two recordings over the same bytes catch state shared beneath them. Over all thirteen
|
|
18
|
+
shapes, for `buildRecordIndex`, `readAnnotations`, `validateRecording`, `readRecords`, `readWindow`
|
|
19
|
+
and `readEnvelope`, plus `inspectEdf` on the bytes.
|
|
20
|
+
- And a third claim the first two cannot make: a second call hands back arrays of its own. Equal
|
|
21
|
+
contents, different objects, different buffers — because two callers reading one recording must not
|
|
22
|
+
be able to write into each other's results. `nothing-points-at-your-buffer` says a result never
|
|
23
|
+
points into the caller's bytes; this says a second result never points into the first one's.
|
|
24
|
+
- `buildRecordIndex` returns methods as well as data, so results are compared through a projection
|
|
25
|
+
that drops function-valued properties — two calls necessarily build two closures, and whether they
|
|
26
|
+
are the same function is not the question. `locate` and `onsetTicks` are then exercised directly at
|
|
27
|
+
every record and at five instants, which is the half that projection drops.
|
|
28
|
+
- Nothing disagreed. No behaviour changed.
|
|
29
|
+
|
|
30
|
+
## 0.6.20
|
|
31
|
+
|
|
32
|
+
- **Added** a thirteenth shape to the `AWKWARD` matrix: a record duration with no exact binary
|
|
33
|
+
form. `formatHeader` carries the story in a comment beside the function it forced — "100 records
|
|
34
|
+
of 0.29 s is exactly 29 s and computes as 28.999999999999996, which floors to 28. The header line
|
|
35
|
+
then reports a recording a whole second shorter than it is (fixed in 0.2.67)."
|
|
36
|
+
- That file is exactly this one, and the matrix had never held it. Every sweep over `AWKWARD` ran
|
|
37
|
+
on durations of 1 s and 0 s — the two values where the float and the rational agree — so the
|
|
38
|
+
arithmetic this whole package is written in ticks to avoid was never exercised by any of them.
|
|
39
|
+
Twenty-three sweeps see it now, and they all pass.
|
|
40
|
+
- `a-duration-float64-cannot-hold.test.ts` spells the hazard out in both directions on the fixture
|
|
41
|
+
itself: the float product is less than 29 and floors to 28, the tick product is exactly
|
|
42
|
+
`29n * TICKS_PER_SECOND`, and the header line reads `00:00:29`. It also checks this is the only
|
|
43
|
+
shape in the matrix whose duration is inexact, and that its derived rate is a clean 100 Hz — an
|
|
44
|
+
inexact duration does not have to make the rate inexact, and a fixture where both were awkward
|
|
45
|
+
would not tell the two apart.
|
|
46
|
+
|
|
9
47
|
## 0.6.19
|
|
10
48
|
|
|
11
49
|
- **Added** a twelfth shape to the `AWKWARD` matrix: a file whose record count field says `-1`, so
|
package/package.json
CHANGED
package/src/constants.ts
CHANGED