eggert at CS.UCLA.EDU
Thu Jun 30 23:09:00 UTC 2005
Arthur David Olson <olsona at lecserver.nci.nih.gov> writes:
> Africa/Cairo: EET-2<EEST>,M4.5.5/0,M9.5.4/24
Are the < and > needed here? I thought the limitations on the
abbreviation length apply regardless of the presence of < and >, and
that the only reason you'd need < and > is if the name contains non
alphabetic characters (are there any such in tz data now? There used
to be some with spaces but I thought we removed them all).
> Europe/London: GMT0BST,M3.5.0/1,M10.5.0/1
> Europe/Paris: CET-1<CEST>,M3.5.0,M10.5.0
These doesn't look right to me. The transition times are always at
01:00 UTC, but this would have the fall transitions at 00:00 UTC.
> $ du -s -k ~/src/tz/tmp/etc/zoneinfo ~/src/tzexp2/tmp/etc/zoneinfo
> 489 /mnt/olsona/src/tz/tmp/etc/zoneinfo
> 779 /mnt/olsona/src/tzexp2/tmp/etc/zoneinfo
I'm a bit puzzled by the growth here. With the POSIX string trick,
only a very few files (Iran, Israel, Chile, Godthab) need 64-bit
entries at all. So shouldn't the overall size of the files be about
the same size as before?
Obviously, as we get closer to 2038 we'll run into more exceptions,
but an advantage of putting 64-bit info only into the troublesome 4
areas now, is that we'll be able to use the poor souls in those 4
areas as guinea pigs.
More information about the tz