Re: [Tzdist] Fun with tzdata and non-Gregorian DST rules

[Paul Eggert <[e-mail]>]

Tim Parenti wrote:
> When it comes time for tz to start
> caring about transitions beyond 2038 (which I'd argue is sooner than
> later!), we may need to take a hard look at natively supporting other
> calendaring systems like RSCALE

Officially time in Iran is astronomical, not arithmetic; so although 
RSCALE-based predictions are more accurate than tz's Gregorian-only rules can 
be, even RSCALE is bound to be wrong at some point. Oscar van Vlijmen's comment 
in the 'asia' file suggests that 2058 is the first problematic year.

As a practical matter it may not help that much to support RSCALE, as the 
Iranian government is likely to change the rules before 2038 anyway.

Come to think of it, the tz entry for Iran can do a better job than it does now; 
proposed patch attached.