chore(release): v1.47.6

This commit is contained in:
Marc-André Bombeck committed 2026-11-15 10:46:35 +01:00
1 parent f2cf316ad8
commit 42c8b11fca
4 files changed
+20 -23

No files matched your search

+17 -20
View File
@@ -42,26 +42,23 @@
links the help page. Brave keeps push messaging off by default, so in
Brave the message names the setting to turn on: "Use Google services
for push messaging" under brave://settings/privacy. (Discussion #1378)
- **Illness episode onset dates keep the intended calendar day.**
Creating a new illness episode in the morning was refused because its noon
default was rejected as in the future, and onset dates could display as the
wrong calendar day in non-UTC time zones. Saving now anchors to the current
time when before local noon and preserves stored timestamps on edit.
- **No more nightly "job still failing" alert over a provider that cannot
answer.** When the only account on an instance still had an AI provider
saved whose key the provider no longer accepted (removing the key from a
Local or OpenAI-compatible provider keeps its address, so it is still
called), or whose provider HealthLog had paused after repeated errors, the
nightly briefing pass counted that account as a failure, and every night
the admins were told the job was broken. A key the provider refuses is now
that account's matter: the run notes it but does not fail over it, unless
it is the instance's own shared key. An account whose providers are all
paused is skipped, since nothing was sent. The Background jobs card on the
admin status page now lists a run that finished with every account
failing, the same runs the alert counts, and the alert and the card say in
words what failed and which provider, instead of quoting the run's
counters. Those counters also stop counting a fallback text as a warmed
assessment. (#1384)
- **Illness episodes can be added in the morning.** The form sent the
start date as noon, which before noon is still in the future, so the
save was refused. Start and end dates now also stay on the day you
picked when your time zone is far from UTC, and editing an episode
keeps the times you saved. (#1376)
- **No nightly "job still failing" alert over an AI provider that cannot
answer.** An account that still had a provider saved whose key the
provider refused (removing the key from a Local or OpenAI-compatible
provider keeps its address, so it is still called), or whose providers
HealthLog had paused after repeated errors, counted as a failed
briefing every night. On an instance with one account that read as a
broken job. A refused key of the account's own is now noted but no
longer fails the run; a refused shared server key still does. An
account whose providers are all paused is skipped, since nothing was
sent. The Background jobs card on the admin status page now lists the
same runs the alert counts, and both say in words what failed and
which provider. (#1384)
### Security
+1 -1
View File
@@ -1,7 +1,7 @@
openapi: 3.1.0
info:
title: HealthLog API
version: 1.47.5
version: 1.47.6
description: >-
Self-hosted personal-health-tracking PWA — public API surface for the iOS native client and external ingest.
+1 -1
View File
@@ -1,6 +1,6 @@
{
"name": "healthlog",
"version": "1.47.5",
"version": "1.47.6",
"description": "Self-hosted personal-health-tracking PWA with Withings integration, AI insights, and doctor-report PDF export.",
"license": "PolyForm-Noncommercial-1.0.0",
"homepage": "https://healthlog.dev",
+1 -1
View File
@@ -16,7 +16,7 @@ try {
}
const CACHE_VERSION =
(typeof self !== "undefined" && self.__APP_VERSION__) ||
/* @sw-version-fallback */ "v1.47.5";
/* @sw-version-fallback */ "v1.47.6";
const STATIC_CACHE = `healthlog-static-${CACHE_VERSION}`;
const PAGE_CACHE = `healthlog-pages-${CACHE_VERSION}`;
// Every HealthLog cache, current and legacy, and nothing another app on the