Hi, please follow also this discussion: (https://github.com/cyrusimap/cyrus-timezones/pull/6):
> elliefm: > I looked deeper into it, and did find and fix one bug in Cyrus, and > added a workaround for libical misbehaving(?) -- details in the issue > I linked earlier. The original issue I thought might exist still might > exist, I haven't tested for it specifically yet due to being > distracted by the stuff I did fix. > > wrt whether or not to packagers should include cyrus-timezones, @rsto > mentioned in our chat: > > > One thing: using the system timezones is not recommended. If I > > recall correctly, some of the system timezone definitions defined > > daylight savings time changes per year, which produced unnecessarily > > long VTIMEZONEs embedded in the VEVENTs returned by Cyrus. > > I have mixed feelings. If, as a mail server admin, you're installing > Cyrus from a system package, it seems reasonable to expect it to use > the timezone definitions provided by the system. It also seems > reasonable that, if you wish to supply your own timezone definitions > instead, that you could configure the zoneinfo_dir setting to point to > them. > > But if, as a package-maintainer, you know your system's timezone > definitions are "basic timezone definitions for general purpose use, > but perhaps inadequate or suboptimal for building a calendaring > service on top of", then it also seems reasonable to package up our > cyrus-timezones package and make the cyrus-imapd package depend on it.
