Conversation
Play rejects the parameter for any edit it has decided must be reviewed: "Changes are sent for review automatically. The query parameter changesNotSentForReview must not be set." The 1.2.7 upload hit exactly that at the commit step, after the AAB had already uploaded — nothing in the workflow had changed since 1.2.6, which committed successfully with the same flag. What changed is the release: 1.2.7 adds FOREGROUND_SERVICE_DATA_SYNC and a dataSync foreground service, and a new foreground service type requires a Play Console declaration and is subject to review, so Play forces the edit through review and refuses to let the flag opt out of it. Holding changes back from review is the wrong default for this app anyway. Reviews here are slow, which is why AGENTS.md tells us to stay ahead of Play's deadlines; a release parked in the console waiting for someone to press "send for review" only adds to the latency it warns about. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The monitor derived its own 15-minute forecast — projected15 = sgv + delta * projectionSteps — and raised "trending high/low" notifications from it. Nothing in the Nightscout status payload carries a prediction; that number was entirely AutoSugar's, which made the app generate new clinical information rather than display the user's own. Prediction and prognosis are named explicitly in the EU MDR's software rule, so it was also the single most medical-device-like thing in an app that is meant to be a viewer for Nightscout. Drop both forecast notification types. The trend arrow and delta already shown on the car screen come from Nightscout itself, so the "where is this heading" information survives without the app forecasting on the user's behalf. Threshold notifications are untouched. Notification ids 1003/1004 are left unused rather than reassigned, so a forecast notification still showing on an older install can never be replaced by an unrelated one. Dropping the forecast also removes the per-check history fetch it needed, so each poll now makes two requests instead of three. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
…lain it getThresholds substituted 70/180 when a user's Nightscout reported no thresholds, and the monitor then notified against them. Those numbers were AutoSugar's own choice of what counts as high or low for someone — the one judgement a display for Nightscout data must not make on a user's behalf. Split the two uses of the values apart. bgLow/bgHigh stay populated for the graph, which needs a scale and a band regardless, but alertLow/alertHigh carry only what Nightscout actually reported and are the only bounds the monitor will act on, per bound rather than all-or-nothing. A source whose Nightscout defines none now stays quiet instead of being measured against a default, and testing the connection says so on the phone rather than leaving the user to infer it from silence. Enabling a source's notifications now opens an explainer first: where the thresholds come from, that the graph only joins real readings with straight lines and forecasts nothing, that stale readings are skipped and repeats are limited, and that AutoSugar is not a medical device. The permission request follows the explanation, which is also the order Android recommends for rationale. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
…pping Two ways this workflow could go wrong quietly. The job checks out the ref it was triggered on but takes its artifacts from the tag, so deploying a tag whose what's-new files are not on that ref made the glob match nothing, the `[ -f ]` guard skip every iteration, and the upload ship with an empty release-notes directory and a green tick. It now counts what it staged and fails with a message naming the version and the ref instead. mapping.txt was also downloaded in the same `gh release download` call as the AAB and passed to the action unconditionally, which made it a hard requirement that release.yml never guarantees — it attaches the file only `if [ -f ]`, so a build without one would have broken the deploy on an asset that is optional by design. It is fetched separately now, its absence is a notice rather than a failure, and mappingFile is left empty so the action skips it. Also correct the character-count instruction in the /release command: it told you to use `wc -m`, which silently counts bytes when the shell has no UTF-8 locale — the default for the non-interactive shells the command runs in on Windows. It reported ja-JP at 383 instead of 137, which would have had someone trimming what's-new text that was nowhere near Play's limit. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.