- The relevant issue here is that there isn’t right or wrong where no clear specification is existing.
- If you think there’s an ISO-approved standard to the effect that the handling of “our case” by the software you named is correct, and the way LibO Calc does it is not, please show me the way to find it.
- Any commercial player in the field surely isn’t authorized to create standards.
- The more general issue is that the common part of the as-if-standard (that’s using a single Double value for a combined representation of date and time) is bad.
- Do you remember the “y2k” crisis? We (“humanity”) should have learnt a lection: Use the needed amount of storage/paper/RAM/whatever for the information you want/need to represent. Never prefer a silly concept of “efficiency” or “de-facto-standards” over reliability and correctness. Otherwise you can’t ever have unambiguous communication - which may be a communication with yourself bridging a few years.
- Once again: Don’t deny the reliabilities of the user.
This isn’t actually a soluition, of course, but having commented badly on the issue in more than one place, I want to apologize and get this clear:
I now think I was wrong concerning the relevant parts of my comments.
The current state of my understanding you find in my comment #4 on the related bug report tdf#139932. Unfortunately I don’t feel capable of analyzing the original code of LibO.