edf2csv 0.5.127 → 0.5.129

package/CHANGELOG.md

@@ -3,6 +3,51 @@
 Notable changes to edf2csv. Versions follow [semantic versioning](https://semver.org); while the
 major version is 0, a minor bump may contain breaking changes.
 
+## 0.5.129
+
+### Fixed: the commonest refusal on the site was quoted with its advice flush left
+
+```
+error: No channel named "ECQ". Did you mean "ECG"?
+Run with --info to list the channels in this file.
+```
+
+The tool indents that second line by seven, under the first word of the message. Four hundred
+lines below its own copy of this block, the reference explains why: "Every refusal takes that
+shape — `error:` on the first line, the advice indented under it — so stderr can be grepped for
+`^error:` and find all of them." Written flush left it reads as a second error, and a reader
+building that grep out of these blocks would expect two lines back from a run that emits one.
+
+The FAQ had it the same way, in the answer titled "I asked for a channel and it says there is
+no channel with that name" — which is the page a reader arrives at holding the real message,
+looking for the one on screen.
+
+Both now match, and a test runs the refusal rather than matching its text: any block on the
+site that opens `error: No channel named "…"` is compared against what that term really
+produces.
+
+## 0.5.128
+
+### Fixed: the step formula the pages print is negative for a calibration real files carry
+
+Three pages state the smallest physical step a channel can express, because it is what decides
+how many decimals its values get. Two printed it with the magnitude on one difference only:
+
+```
+step = |physical_max - physical_min| / (digital_max - digital_min)
+```
+
+`reversed-bounds.edf` has a channel whose digital pair is written the wrong way round — legal,
+warned about with `INVERTED_PHYSICAL_RANGE`, and converted as the header says. For that channel
+the printed formula is -0.1, and the next thing the page tells you to do with it is take
+`ceil(-log10(step))`, which of a negative number is not a number at all. The code takes the
+magnitude of the whole quotient and gives that channel 3 decimals, the same as the upright
+channel beside it — the only answer that keeps its distinct codes distinguishable.
+
+Both differences are magnitudes now, on all three pages, with a sentence saying why. Held by a
+test that evaluates the formula the page prints against the function the conversion uses, on
+each of that fixture's three shapes: physical pair reversed, digital pair reversed, and both.
+
 ## 0.5.127
 
 ### Fixed: the chunked-reader recipe printed a row count and a peak from nowhere

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "edf2csv",
-  "version": "0.5.127",
+  "version": "0.5.129",
   "description": "Convert EDF, EDF+ and BDF biosignal recordings (European Data Format) to CSV from the command line. Local, streaming, and never resamples or alters units.",
   "keywords": [
     "edf",