getChatzos does not return null when sunrise is null

[MichaelJBerk]

I have been testing KosherJava against some of the tests in Moshe Dicker's rust port (specifically, the Polar Day test ).
I've found that when running getChatzos using the parameters used in the location day test, the method returns a value, rather than the expected result null. Other methods on the ZmanimCalendar seem to return null on that day as expected.

To recreate this issue:

double lat = 69.6492;
double lon = 18.9553;
double elev = 0;
TimeZone tz = TimeZone.getTimeZone("Europe/Oslo");

GeoLocation loc = new GeoLocation("Polar Location", lat, lon, elev, tz);
ComprehensiveZmanimCalendar czc = new ComprehensiveZmanimCalendar(loc);
czc.getCalendar().set(2017, 6, 21);
System.out.println(czc.getChatzos().toString());

[KosherJava]

@MichaelJBerk ,
This is an interesting question. @dickermoshe can comment on the Rust implementation, but returning a non-null in the Arctic where there are no other zmanim is indeed possible, and actually required by some poskim in some situations. First have a look at my article The Definition of Chatzos. If you look at Zmanimcalendar.getChatzos(), you will see that if (isUseAstronomicalChatzos()) { (it defaults to true) it will return getSunTransit(). If the NOAACalculator is in use (the default), this method calls the NOAACalculator.getUTCNoon(LocalDate, GeoLocation). This calculates chatzos not as half a day, but as true solar noon that exists even in the Arctic (it only uses the longitude and not latitude as part of the calculation). Even if changed to false, if noon as half a day returns null, it will return astronomical chatzos. If you change to the less accurate older USNOCalculator, you will generate a null, since that calculator calculates chatzos as half a day from sunrise to sunset.
I mentioned poskim discussing this, and a concrete example is Shabbos and other zmanim in the Arctic circle. Rav Moshe Sternbuch's psak in such a case is to consider chatzos halayla, when the sun is at its lowest position in the sky as the beginning of Shabbos (it can't get any lower, so that is the definition of sunset). This can is calculated using true solar midnight when the sun is at its lowest point along the horizon (directly north of the observer in the northern hemisphere). There are other poskim who seem to hold of variations on this concept, including the Ben Ish Chai (quoting the Divrei Yosef).
In short, by default, true solar noon will be returned, and not the null you expected.
Note - this reminded me that I have to create a getChatzosHalayla() wrapper, that behaves the same way.
I have a long article titled "Calculating the Day-Night Boundary in the Land of the Midnight Sun" in draft that I hope to eventually publish. That covers the subject in detail.

[MichaelJBerk]

I've read through the article and I see why there would still be a Chatzos in the article circle at that time. Would it make sense to update the documentation to reference this?

[KosherJava]

@MichaelJBerk , Have a look at the ZmanimCalendar docs where as far as I can see, all methods link to that article. The methods in the parent AstronomicalCalendar also reference the article. If you are aware of any location where the documentation can be improved, please let me know.
I appreciate all and any testing (such testing has exposed issues in the past, and I really appreciate anyone who takes the time to run tests), please please do not hesitate to continue providing feedback.
In addition, please feel free to submit PRs with test cases. There is a directory of test cases, and I would love to have them expanded.