The current plan for Atlas's calendar system consists of the following:
a) the user may choose to use the Gregorian calendar, OR:
b) the user may specify their own calendar with the following features:
  - a single repeating week,
  - the names of the days of the weeks,
  - a series of months (each with names and lengths),
  - years are assembled by specifying a list of months (e.g. for Gregorian, there are two years, distinguished by which version of February they use)
  - a year cycle, consisting of several years (e.g. for Gregorian, 3 normal years followed by 1 leap year)
  - named years and eras
  - special "one-shot" years or rare recurring years—this would be a uniform way of implementing both intercalation and skipped leap years
  - as many independent calendars as you want, for e.g. religious calendars that are separate from natural calendars, or for handling changes in calendar systems
  - holidays based on a date (e.g. "dec. 25th"), a week day within a month (e.g. "first tuesday in smarch"), or a weekday within a month relative to another holiday (e.g. "first tuesday in april after the new moon", where "the new moon" is a recurring event defined using a separate lunar calendar)

This leaves me with the following questions:

1. Does anyone use the Julian calendar on their alt-earths (or elsewhere)? What about other real-world calendars? (These could be implemented directly just like the Gregorian calendar.)
2. Does anyone use a non-repeating series of year lengths and month configurations? (If so, I hate you, but I'll do what I can.)
3. Does anyone have a con-calendar with a realistic leap year correction (e.g. "skip a leap year every 400 years" and other obscurities of the Gregorian system)?
4. What kinds of nasty holiday date calculations do you use? Anything worse than Easter?

When this is all done, you'll be able to specify dates for events (with any specificity) and see them assigned appropriately to a well-structured, good-looking calendar chart, as well as recurring holidays, so working out the logic for this is important.

The traditional Japanese calendar? If I ever work on that Ralts-line conlang I often mention, they'll use that.

Well, Êndug's chronology is currently based on Ab Urbe Conditem (AUC/AVC), and while the timeline got seriously perturbed, Gaius Julius still rose to power, so he definitely put through the Julian reform before being demised (teaser quote from one of my timeline blurbs in 548AD "...the General, Marcus Antonius Severus (descendant of the great Triumvir who helped lead the Republic through one of its most trying moments some 500 years before) consulted with his advisors...").

One should be able to specify sequences of days, weeks, months, years, and sets of holidays, but be able to have these sequences change at different points. This is important because, in Real Life, rulers often arbitrarily changed calendars at points, especially in ancient times but as recent as in early modern times, and yet one would want to be able to arbitrarily change the calendar however one wants.

One should also be able to have time that is relative to a given "era", where the era may change, sometimes frequently (frequently with changes in rulers). The reason for this is that such eras were the norm for calendars in East Asia up until pretty recently, all things considered.

quoting Halian:

The traditional Japanese calendar? If I ever work on that Ralts-line conlang I often mention, they'll use that.

...I'm not sure what this is supposed to mean exactly; Japan used Chinese-derived systems until consistent contact with the west and then adopted the Gregorian calendar plus 660 years (kōki/kigen). Did you have some specific system in mind?

I expect the system will be sophisticated enough to support specifying the whole Gregorian calendar, as well as intercalation (blame pthag), so there shouldn't be an issue with kōki.

I will, however, make a point of supporting named years and eras, since that seems to be important.

quoting Travis B.:

One should be able to specify sequences of days, weeks, months, years, and sets of holidays, but be able to have these sequences change at different points. This is important because, in Real Life, rulers often arbitrarily changed calendars at points, especially in ancient times but as recent as in early modern times, and yet one would want to be able to arbitrarily change the calendar however one wants.

Holidays will be flexible, but not the whole calendar system (except for intercalation and leap years). If you want to do a big switch, e.g. from Julian to Gregorian, then you'll create a separate calendar. Otherwise that's just ridiculous amounts of bloat.

One should also be able to have time that is relative to a given "era", where the era may change, sometimes frequently (frequently with changes in rulers). The reason for this is that such eras were the norm for calendars in East Asia up until pretty recently, all things considered.

Will do, along with named years.

quoting Rhetorica:

quoting Travis B.:

One should be able to specify sequences of days, weeks, months, years, and sets of holidays, but be able to have these sequences change at different points. This is important because, in Real Life, rulers often arbitrarily changed calendars at points, especially in ancient times but as recent as in early modern times, and yet one would want to be able to arbitrarily change the calendar however one wants.

Holidays will be flexible, but not the whole calendar system (except for intercalation and leap years). If you want to do a big switch, e.g. from Julian to Gregorian, then you'll create a separate calendar. Otherwise that's just ridiculous amounts of bloat.

It may seem like bloat, but it is necessary because this is how calendars operated in premodern times, at least until the likes of the Julian and Islamic calendars were adopted. So if one wants to set one's calendar in premodern times, realistically, it is important to be able to support this.

I should work the leap year thing for mine... I was tempted to base it on the Solar Hijri algorithmic calendar. On the other hand I know some holidays, but I should work more on it, so I have the full list.

Hmm... Guess we're gonna need to be able to insert/delete special years, too. Oi moi. Okay.

Question: is the calendar system just a timeline thing for the entire world or is the idea for people to be able to reflect the time structure of individual cultures? I ask because reflecting the calendars of premodern cultures includes both things like regnal years and weekless calendars as well as really silly things à la the Mayan calendars or non-arithmetic ones (there was, for instance, a Salish tribe where the new year started with the first spawning of Chinook salmon; they would then count a number of months and then stop counting until the salmon run).

The primary purpose of the calendar system is for keeping track of the world's timeline, but to do so in as con-culture-y a way as reasonably possible. That won't really be compatible with discontinuous calendars, although I'm sure you could find a way to simulate such a thing just by marking dates on a continuous calendar. (It's good to know that kind of thing exists, though!)

Systems with no week won't be a problem—hopefully the core features of the Mayan system will be fully implementable.

Some day, some day... I will get to this, once my academic workload quiets down. There seemed to be more interest in the automatic morphology and diachronics stuff, though, so that's going to come first—again once academia regresses and Morrígan has gotten a bit more time to play with her SCA s'more.

Meep? Not trying to be pushy or anything but this would surely be quite a welcome addition.

(edit: I can't count, it looks like it IS 365 days).

Saltha's calendar has 14 months each with 5 weeks and each week with 5 days, and then a 15th month with 15 days that are out of the year. So basicly being able to set # of days in a week would make this work. Also, setting 1st/last day of the year so it doesn't have to be Jan 1st.

quoting Rhetorica, Kelatetía:

quoting Halian:

The traditional Japanese calendar? If I ever work on that Ralts-line conlang I often mention, they'll use that.

...I'm not sure what this is supposed to mean exactly; Japan used Chinese-derived systems until consistent contact with the west and then adopted the Gregorian calendar plus 660 years (kōki/kigen). Did you have some specific system in mind?

I expect the system will be sophisticated enough to support specifying the whole Gregorian calendar, as well as intercalation (blame pthag), so there shouldn't be an issue with kōki.

I will, however, make a point of supporting named years and eras, since that seems to be important.

The former, actually, month lengths that literally vary with the phases of the moon and all.