libmuslim 0.1.1

Dart FFI bindings for libmuslim: prayer times and other tools for Muslim applications.

0.1.1

Vendored header

Synced to prayertimes.h v0.2.3, libmuslim releases 2026.08.21 and 2026.08.22. No API change: every type and member in this package is unchanged, and code written against 0.1.0 compiles against this release.

refine_event is gone (libmuslim#85, closing libmuslim#79). Fajr, isha, sunrise and sunset are now solved the same way, by bisecting true solar altitude between local noon and local midnight, so whether an event exists is decided at the event rather than at 0h UT. The twilight grazing residual that motivated the change is gone with it: the worst angular error over the flagged points falls from 30.1964 arcmin to 0.1970.

Times move at high latitude, and what you see depends on the method. Measured directly against the 2026.08.21 header over latitude -70 to 70 in steps of 5, longitude 0, every day of 2026, all five prayers, 52925 values:

	MWL	Kemenag
values changing the displayed minute	14	13
prayers becoming unavailable	0	3

Every changed value is a day on which the twilight event does not exist and the old solver reported a crossing near midnight anyway. Under a method that publishes a high-latitude rule, MWL and Moonsighting, the substitution now runs instead, which moves isha by up to 1 hour 54 minutes and never removes a time. Under the other 20 methods there is no rule to fall back on, so PrayerTimesUnavailable is thrown where 0.1.0 returned a wrong time. On this grid that is latitude 70 on 2026-05-16, where Kemenag now reports [Prayer.fajr, Prayer.maghrib, Prayer.isha] as unavailable and 0.1.0 returned 00:07, 23:33 and 23:49.

Every other value on the grid moves by at most 0.0003 seconds, far below the whole-minute resolution this package returns.

Existence was then checked upstream against JPL DE440 over the same grid, 42340 decisions across sunrise, sunset, fajr and isha. Two disagree, both a single day at latitude -70 in November where the low precision solar model and DE440 differ by 0.0035 degrees on a 0.0013 degree margin. That is the sun model's own noise rather than the solver's.

0.1.0

First published release. Versions 0.0.1 and the entries previously filed under Unreleased were never published to pub.dev; their content is folded in here.

• Dart FFI bindings for prayertimes.h from libmuslim. The header is vendored at src/prayertimes.h and its ABI is pinned by test/prayertimes_abi_test.dart, so a changed C layout fails the tests rather than diverging silently.
• PrayerTimes returning whole-minute UTC instants, with timeOf, current, next and timeUntilNext. It carries the five prescribed prayers only. Sunrise is the end of the fajr window rather than a prayer, and dhuha is a voluntary prayer carried only by Indonesian timetables. Both are still computed inside the C library, because maghrib is sunset and every high-latitude substitution measures the night between sunset and sunrise, but neither is part of the contract (libmuslim#63). Prayer.values is therefore exactly the set of prescribed prayers.
• CalculationMethod and CalculationParameters, including custom angles, the hanafi asr school and an ihtiyat override.
• HighLatitudeRule, plus highLatitudeRule and highLatitudeReferenceLatitude on both CalculationParameters constructors. Inside the polar circle there is no sunrise or sunset, so the substitution rules have no night to measure, and only MWL and Moonsighting carry a reference latitude. A caller serving a location the chosen authority is silent about can supply one themselves, which keeps the choice in the caller's code rather than misattributing it to that authority (libmuslim#51).
• Prayer and PrayerTimesUnavailable.

Vendored header

Synced to prayertimes.h v0.2.1, libmuslim release 2026.08.20.

Prayer times are computed from a solar position evaluated at each event's own instant. Maghrib is measured against a JPL DE440 validated solver at 6.5966 seconds worst case for |latitude| <= 60. Fajr and isha have no oracle behind them upstream, tracked as libmuslim#52.

Inside the polar circle the whole day is solved at the reference latitude rather than borrowing only sunrise and sunset. Each value in the tests is checked against the C library directly.

Known issue

PrayerTimes.forDate throws PrayerTimesUnavailable with [Prayer.asr] on a narrow band of days at high latitude. At Longyearbyen that is four days a year, where the separation from the declination sits between 90 and 90.833 degrees: the Sun is visible only by refraction, so sunrise exists and fajr, maghrib and isha all resolve, but nothing casts a shadow. Four valid times are withheld with the impossible one. That is the same shape as the dhuha problem libmuslim#63 solved upstream by removing the field; asr cannot be removed, so whether it should become nullable is open.