Changelog for swiss-ephemeris

v1.4.2.0 (2021-12-03)

• Fix edge case in grav group: incorrect casting in the C code was causing planets that were too close to a sector boundary or another planet to be thrown back into the first sector.
• Remove cuspsToSectors.
• Add ability to set longitude in HasEclipticLongitude typeclass

v1.4.1.0 (2021-11-27)

• Export utcToJulianDays, to obtain a product of (TT, UT1) Julian Days from a UTCTime value -- saves you one IO trip vs. getting them separately.
• Support for GHC 9.2.1
• Fix minor bug in directionChange that made lookups starting <=30 minutes before the event fail (due to an artifact of the original C function that explicitly excludes such lookup bounds.)

v1.4.0.0 (2021-11-11)

BREAKING CHANGE: Major refactoring of time values: JulianDay is no longer an alias for Double, and is now a type that carries a witness of its provenance -- which enables us to work with functions that transact in both Terrestrial (Ephemeris) Time, and Universal Time, without mixing them up.

• Upgrades to version 2.10.02 of the C library.
• Adds the Precalculated namespace with functions to read and write pre-calculated ephemeris from a file on disk. Useful when examining an interval of time for ecliptic phenomena.
• Introduces the SwissEphemeris.Time module, with various conversions between time standards and some Haskell time values.
• Introduces functions that find moments of exactitude for longitude crossings (heliocentric and geocentric,) and moments of exactitude for lunar phases. The geocentric interpolation, as well as lunar phases, use the Brent-Dekker algorithm for root finding, the heliocentric functions are able to do faster parabolic approximation. I need to study the C sources a bit more to see if something similar is possible for interpolation in general (see the events files in the C sources for ideas.)
• Adds functions for calculating planetary phenomena as visible from earth, as well as solar and lunar eclipses.
• [dev] adds a Dockerfile and NOTES.md to aid in diagnosing memory leaks.
• [dev] adds some basic Nix derivations for producing documentations and a release tarball.

v1.3.0.2

• More memory safety paranoia: ensure that gravGroup functions have appropriately scoped "extraData" pointers; was using an unsafe function before that somewhat inexplicably never failed, except for one time maybe while doing unrelated testing.
• Fix empty sector handling for gravGroup: it now correctly reports that zero-length sectors are not supported.

v1.3.0.1

A couple of memory safety patches:

• Attempt to rein in memory unsafety by keeping all pointer peeking in IO for gravGroup fns.
• Always allocate 256 chars for error messages.
• [dev] Bundle test ephemeris into the hackage tarball, to allow hackage CI and nixOS to successfully run tests.

v1.3.0.0 (2021-06-18)

• Drops support for base < 4.10, which effectively excludes GHC versions less than or equal to 8.0.2. Supporting older haskells gets more cumbersome with each addition. Please submit a patch with all the preprocessor magic if you really want old haskell support.
• Add ChartUtils namespace, with a couple of convenience functions for chart drawing.
• Upgrade to swiss ephemeris 2.10.01.
• Update LICENSE to AGPL -- it changed in the C library, too.
• [dev] Fix QuickCheck test dependency, to hopefully be copacetic in NixOS builds.
• [dev] add optional nix scripts.
• [dev] a whole buncha hlint + autoformat fixes.

v1.2.1.1 (2021-03-14)

• Bump upper bound for base, to work with newer haskells.

v1.2.1.0 (2020-11-14)

• Introduces gregorianDateTime, which is the reverse of julianDay: given a JD, return a tuple with (year, month, day, decimalHour).

v1.2.0.0 (2020-09-14)

BREAKING CHANGE: splitDegrees now takes options that reflect the options in the underlying library.

• Constructors for JulianTime and SiderealTime are now exposed.
• Introduce SplitDegreesOption enum for all options one can split degrees with; amends splitDegrees to take said options as the first argument.
• splitDegreesZodiac is unchanged, though a mere veneer for the now more powerful splitDegrees.
• Since splitDegreesZodiac goes the extra enum-mile to provide human-readable zodiac names, and the underlying library can also split on Nakshatras, we now include the NakshatraName enum. Names are from wikipedia and I saw some variants, so please forgive any mispellings!

v1.1.0.0 (2020-09-12)

BREAKING CHANGE: the Coordinates type has been retired, in favor of the more specific GeographicPosition and EclipticPosition. calculateCoordinates is now calculateEclipticPosition, and the calculateCusps* family now takes a GeographicPosition as part of its inputs.

• Introduces an Internal module with types and helpers that this library introduces, which are not native to the underlying C library. Import at your own risk! (the "curated"/ "stable" ones are re-exported by the main module.)
  • Deprecates the Coordinates type, in favor of EclipticPosition.
• Introduces functions to calculateEquatorialPosition and calculateObliquity at a given time, as well as types that better convey the different types of positions (EquatorialPosition, ObliquityInformation).
• Some astrology helpers: convert between equatorial and ecliptic (and vice-versa,) obtain the Delta Time effective at a given moment, obtain the house position of a given body. (Note: the calculateHousePosition function is more useful for working near the polar circles or for bodies off of the ecliptic -- the ARMC and obliquity need to be calculated or provided, it's simpler if you already have the cusps: just check which cusps a given longitude falls between -- no need for this function!)

v1.0.0.0 (2020-09-07)

• Refactor the calculateCusps function:
  • Return a simple list of cusps. This allows for future implementations of exotic systems that have more (or fewer?) cusps, and hews closer to regular usage (which iterates over the cusps.)
  • The house system comes first, to allow for more ergonomic partial application for uses where one system is the "default" (e.g. traditionalCusps = calculateCusps Placidus.)
• Cleans up haddocks, adds many links to the original docs (and notes the headings, since updates to those seem to break hyperlinking?)

v0.3.1.0

• Fixes occasional segmentation fault (caught most often in the more memory-strapped CI server than in my computer,) caused by using alloca for the error string and, when no error string was populated, ending with undefined behavior. Now we explicitly allocate the 256 chars that the documentation and C sources recommend, which seems to be always freed by Haskell, vs. leaving a hole somewhere when the underlying library fails to terminate the string.

v0.3.0.0

Breaking fixes to calculateCusps and calculateCoordinates

• Upgrades to v2.09.03 of the C library, to incorporate some bug fixes that seem marginally related to random breakage I've seen; read more at: https://www.astro.com/swisseph/swephprg.htm#_Toc49847971
• Introduces withoutEphemerides which sets the ephe path to NULL (via the also new setNoEphemeridesPath) and takes care of calling closeEphemerides. Use this or withEphemerides for memory safety, only call the functions directly if you really know what you're doing (i.e setting/closing ephemerides in some other manner.)
• Both calculation functions are now IO computations, to reflect the fact that they may interact with ephemeris data and allocate memory that closeEphemerides has to free.
• More closely reflects the underlying behavior for calculating cusps: it may return cusps in the Porphyrius system if given a point for which the chosen system fails. To more explicitly reflect this, we now have calculateCuspsStrict which returns a Left value if the requested house system couldn't be used. calculateCuspsLenient always returns a calculation, and is aliased to calculateCusps as the "default" method.
• Since the calculation may have changed the house system, we now return a systemUsed entry in the CuspsCalculation record.

v0.2.0.0

• Introduces withEphemerides for bracketed access to the ephemeris directory.
• Changes the signature of calculateCusps to return a Left value if the underlying library is unable to calculate the cusps.
• Introduces "monadic" versions of the calculations that work with instances of MonadFail: calculateCuspsM and calculateCoordinatesM
• Improves test coverage with property testing.

v0.1.0.0 - 0.1.0.2(2020-08-12)

• Bundles the C code for v2.09.01 of Swiss Ephemerides -- refer to that page and related documentation for other possible additions to this package!
• First release with basic bindings to calculate the coordinates of bodies between the Sun and Chiron, plus cusps and major angles -- calculateCoordinates and calculateCusps, respectively.
• The functions setEphemeridesPath and closeEphemerides are provided to initialize (important) and release (less important) resources related to caching calculations and locating the data files for ephemerides.