An offline sun time calculator
------------------------------
On the final day of eligibility, here's a quick post introducing
my project for OFFLFIRSOCH 2025. I kind of liked the extremely
understated and utilitarian feel of naming my 2024 project `city`,
and so I stuck to that theme and this year's entry is called `day`[1].
`day` is a simple tool for calculating when the sun rises, reaches
it's highest point in the sky (solar noon), and sets in your location
and in your local time, as well as telling you how long the day is
(defined here as the time between sunrise and sunset). It can,
optionally, tell you a few extra things, like the times when the
three different kinds of twilight (civil, nautical and astronomical)
begin and end, and the duration of "usable light" in the day (you
can still see well enough to get around and do things outside without
artificial light for a little while before/after the sun itself has
become (in)visible), how many degrees above the horizon the sun it
at its highest point that day, as well as the solstice and equinox
dates for the year. By default it shows you all this information
for today, but you can use command line arguments like +5d or -1m to
calculate it for, e.g., 5 days in the future or 1 month in the past.
In order for this to work, you need to provide your latitude and
longitude via an environment variable, $LATLON. If you don't
know your latitude and longitude you could, of course, use `city`
to find a usable approximation.
The whole thing is less than 500 lines of Lua, although I do rely on
a third party library, luatz, to take care of timezone conversions
and account for DST changes and that kind of mucky stuff. The vast
majority of the formulae used for these calculations were taken
from a hard copy of a 45 year old book (Jean Meeus' "Astronomical
Formulae for Calculators") which I bought off eBay for 10 EUR.
That is the sum total of what's necessary to replace the entire
"Sunrise & Sunset" section of timeanddate.com (who, I kid you not,
will tell you sun and moon related times down to the second, rather
than the minute, if you give them 15 USD per year!), or an even
larger portion of the functionality of e.g. sunrise-and-sunset.com.
While debugging `day` (ironically I relied on Wikipedia instead of
Meeus in the beginning and was getting bogus results), I at one
point double checked my conversions from the Gregorian calendar
to Julian Dates (widely used in astronomical calculations) against
those at www.juliandate.org. Here's what that site wants to feed
you cookies for:
* Store and/or access information on a device (769 partners included)
* Develop and improve services (399 partners included)
* Use limited data to select content (113 partners included)
* Use limited data to select advertising (507 partners included)
* Create profiles for personalised advertising (570 partners
included)
* Use profiles to select personalised advertising (565 partners
included)
* Create profiles to personalise content (249 partners included)
* Use profiles to select personalised content (222 partners included)
* Measure advertising performance (550 partners included)
* Measure content performance (288 partners included)
* Understand audiences through statistics or combinations of data
from different sources (397 partners included)
The function in `day` which provides precisely the same answer
as that site is less than 10 lines of code. The ratio of cookie
"partners" to lines of equivalent local code here is at least a
double digit number, quite possibly a three digit one. This is
the world we live in, folks!
There's more I'd like to say and maybe I'll do a follow up post,
but unfortunately I don't have much spare time this evening, so
this will have to do. If `day` is of interest to you, please give
it a whirl and let me know about bugs or glaring usability defects.
I am sure they exist! If it predicts sunrise or sunset times for
your location which differ by more than, say, 3 minutes either way
from some other site or app you use, please let me know. If it is
off by 1 minute or 2 minutes I am not really bothered - something
I realised early on in this project is that any two seemingly
trustworthy sources for this information will in fact always
disagree on at least one quantity by a small margin. There's just
no way around it. Most of them do not make their code available or
explain what methods they use so there is no way to understand why.
This is also the world we live in. At the end of the day, I actually
trust `day` more than anything else because I know for a fact it
at least tries to implement the formula in a book published by a
university and written by someone with a degree in mathematics,
and that's more than I can say for *any* website.
Let this post also serve as a reminder that you need to announce
your OFFLFIRSOCH project via Gopher and/or Gemini before the end
of March (in your timezone) to qualify. The clock is ticking by
this point! So far I am aware of only two other entries (sorry,
you two, that I haven't replied yet!)...
[1]
https://git.sr.ht/~solderpunk/day
