Re: calendar gets wrong end for Daylight Savings Time

[Glenn Morris]

Richard Stallman wrote:

> It seems that the Emacs calendar should have a data base of history
> of daylight savings time for various named regions. But maybe that
> is too much work to be feasible.

One could hard-code a few special rules, but I don't think doing it
properly is feasible, no. Presumably all this information is in the
system library timezone database in some format, so rather than
recreating that I'd be inclined to try to extract it from there first.
But I have no idea if that's practical, especially on non-Unix
systems.

Instead what I am considering is checking for the dates of the summer
time transition in every year as and when necessary (eg to work out
the holidays in 2004, you need to check for the transition dates in
2004). This would obviously be slower than the current assumption of
the same value for every year, so I'd: a) make it optional (but
probably the default, so long as it is not too slow); and b) cache the
results for each year so they could be reused.

I don't know if it will work in practice, I'll give it a try and
see...