12 days issues

[kuzmoyev]

Related to #111

1. "Points per day" and "Leaderboard (stars)" have 31 days (x-axis). (Other two charts have correctly 12 days)

2. In "Delta-focused stats" -> "Overview", the "stars" column shows progress as a background of the cell. But it is calculated from 25 days, not 12.

Side-note: thank you for this great tool! If I have time I'll look into opening a PR for this issue.

[jeroenheijmans]

Oh hah, yeah thx for the report, seems we missed some edge cases!

Note that in a handful of places the "31 days" is intentional, even for these 12-day events. But the one you screenshoted is not, that should cap at 12 days for sure!

I'll be working on my survey results next 2-3 days first though, before getting to addon changes. Then again, I hope we get a 12-day 2026 AoC too, so that would make fixes worthwhile for next year too!

[kuzmoyev]

Another note. I realised that some people may solve the problems later, after the 12th day, so the graphs can have data outside of these 12 days.

I'm not sure what's the best way to handle this. In theory, it can be any length on x-axis, in practise one month would be enough. I'd vote for trimming the x-axis to the data and one month, i.e.:

start = 1
end = max(latest-open-problem-day, min(31, latest-leaderboard-submission))

[jeroenheijmans]

I'm not sure what's the best way to handle this. In theory, it can be any length on x-axis, in practise one month would be enough. I'd vote for trimming the x-axis to the data and one month

Could you elaborate? Or, put differently: isn't that what is currently happening? The x-axis is now fixed to 31 days for option 1 ("POINTS"), the whole of December. Your screenshot shows that too, and here's one from an earlier year where you can see who "won" our leaderboard AoC during December (and it gives a hint who completed it some time later):

For the other two options ("PERCENTAGE", and "WITH POTENTIAL"), a cut off of 12 days is better since the x-axis refers to "puzzle days", not calendar dates.

And perhaps all this means the graphs should be split up into at least 2 or maybe even 3?

Summarizing:

1. POINTS variant => x-axis is continuous, referring to timestamps
2. PERCENTAGE variant => x-axis is ordinal, referring to puzzle days
3. WITH POTENTIAL variant => x-axis is ordinal, referring to puzzle days

Let me know if you see this the same, and if you agree with making 2 & 3 capped at 12 for x-axis in 2025+ years.