Inconsistent months data for date conversion

[aj3sh]

We have noticed inconsistencies in the monthly data across Hamro Patro and Nepali Patro, starting from the year 2081. This inconsistency was initially highlighted in the go-nepali project (opensource-nepal/go-nepali#15). The issue was addressed in this project through PR #79. However, this PR only fixes the data for the year 2081.

After a thorough examination, here's what I discovered:

• Hamropatro has incorrect data for the year 2087, with 367 days in a year.
• The data pattern of the Baishakh from Nepali Patro appears to follow a specific pattern and is likely to provide accurate information.

Here are the data from different sources:

[Year]
[Current Data]
[Hamro Patro Data]
[Nepali Patro Data]

[2082]
30 32 31 32 31 30 30 30 29 30 30 30
30 32 31 32 31 30 30 30 29 30 30 30
31 31 31 32 31 31 30 29 30 29 30 30

[2083]
31 31 32 31 31 30 30 30 29 30 30 30
31 31 32 31 31 30 30 30 29 30 30 30
31 31 32 31 31 31 30 29 30 29 30 30

[2084]
31 31 32 31 31 30 30 30 29 30 30 30
31 31 32 31 31 30 30 30 29 30 30 30
31 32 31 32 31 30 30 30 29 29 30 31

[2085]
31 32 31 32 30 31 30 30 29 30 30 30
31 32 31 32 30 31 30 30 29 30 30 30
30 32 31 32 31 30 30 30 29 30 29 31

[2086]
30 32 31 32 31 30 30 30 29 30 30 30
30 32 31 32 31 30 30 30 29 30 30 30
31 31 32 31 31 31 30 29 30 29 30 30

[2087]
31 31 32 31 31 31 30 30 29 30 30 30
31 31 32 31 31 31 30 30 30 30 30 30
31 31 32 32 31 30 30 29 30 29 30 30

[2088]
30 31 32 32 30 31 30 30 29 30 30 30
30 31 32 32 30 31 30 30 29 30 30 30
31 32 31 32 31 30 30 30 29 29 30 31

Possible solutions

1. Align with one of the data sources.
2. Wait for the updated 2082 data and compare which source had the correct data.
3. ...

[khumnath]

@aj3sh i have created those data calculation algorithm in https://github.com/opensource-nepal/nepdate-cli/tree/main/src bikram.h and used in main.cpp. you can check npm package bikram-js or https://github.com/build-for-nepal/bikram-calendar for typescript implementation.

[aj3sh]

@aj3sh i have created those data calculation algorithm in https://github.com/opensource-nepal/nepdate-cli/tree/main/src bikram.h and used in main.cpp. you can check npm package bikram-js or https://github.com/build-for-nepal/bikram-calendar for typescript implementation.

Wow thats great. Will look into it. :-) Thanks for sharing.

[aj3sh]

@khumnath, Your algorithm has 32 days in Ashar 2082, but Hamropatro and Nepalipatro both have 31 days. Do you have any idea about that?

[aj3sh]

The data for the year 2082 on Hamropatro has been updated and now aligns with the data on Nepalipatro, as opposed to the data previously mentioned in the issue. However, Hamropatro still contains incorrect data for the year 2087, showing 367 days in a year.

Therefore, the current data will be updated using the Nepalipatro source.

[khumnath]

i think month index should change. every 32 days in a month should in shravan. let me see if i got time

[khumnath]

@aj3sh If the past dates (for which we have data) align correctly with my calculations, future dates may still have incorrect data on Nepali Patro or Hamro Patro. i have tested only some birthdays and aligned current month with weekday name.

[aj3sh]

@khumnath I will check if your data matches with the past data.

[aj3sh]

@khumnath I have cross-checked your data with nepali-datetime (our package), Hamro Patro, and Nepali Patro. There are 121 mismatches from the years 2000 to 2081. Here are a few sample dates. Please review and verify.

Year	Month	nepali-datetime	bikram-js	nepalipatro	hamropatro
2000	1	30	31	30	-
2001	9	30	29	30	30
2001	10	29	30	29	29
2002	4	32	31	32	32
2008	1	31	30	31	31
2008	2	31	32	31	31
2008	3	31	32	31	31
2008	4	32	31	32	32
2008	10	29	30	29	29
2020	3	31	32	31	31
2020	9	30	29	30	30
2024	3	31	32	31	31
2024	10	29	30	29	29
2026	12	31	30	31	31
2039	4	32	31	32	32
2039	11	30	29	30	30
2043	4	32	31	32	32
2043	9	30	29	30	30
2051	3	31	32	31	31
2051	4	32	31	32	32
2051	10	29	30	29	29
2058	1	30	31	30	30
2062	3	31	32	31	31
2062	6	31	30	31	31
2070	3	31	32	31	31
2076	9	29	30	29	29
2078	9	30	29	30	30

[khumnath]

I think the calculation for the first day of the solar month should be re-evaluated here: https://github.com/build-for-nepal/bikram-calendar/blob/main/src/index.ts#L90-L115. First, we need to research whether the Bikram Sambat in Nepal uses the Drik Siddhanta or Surya Siddhanta method for its calculations.

[sumanchapai]

These are my questions, concerns before approving the PRs on the repos that I have been requested:

@aj3sh , you have highlighted inconsistencies between hamropatro, nepalipatro and bikraj-js #82 (comment). Upon closer inspection it looks like hamropatro and nepalipatro are consistent in the table. Are there inconsistencies between the two for future date, like 2087 that you mentioned? Which source are we using as authoritative for the PRs opensource-nepal/go-nepali#26 (comment) and opensource-nepal/py-nepali#80 (comment)?

On a somewhat higher level, could you please share your perspective on why making the BS calendar has been so challenging? We should have always used a library that algorithmically calculates the days in a month, but that seems to be challenging to build. Every year when the joytis come up with the calendar, what algorithm do they use? Work of @khumnath seems to make a way for this. But without knowing the algorithm ourselves, we cannot verify or change the code. Can you tell us more about how this stuff works? Is there a blog or a video? I have seen that you've used magic numbers such as source:

  double YugaRotation_star = 1582237828;
  double YugaRotation_sun = 4320000;

Would a library like astropy help us in any way, for example knowing the positions of the planets and the stars in the past/present/future?