On Wed, 2003-08-27 at 01:15, William Bloom wrote: > I finally tracked the timezone offset shifting problem down to > icaltime.c I am working with version 1.4.4 sources. The problem is as > follows... > > When the summary view is selected (sometimes other views as well, I > suspect) some steps are done that require temporarily setting the > timezone to UTC. So set_tz() is called to change the TZ environment > variable to "TZ=UTC" using setenv(), then unset_tz() is called to put > the TZ environment variable back the way it was afterward. Now, in > FreeBSD, the TZ environment variable usually doesn't exist because > tzset() typically just works from /etc/localtime. In this case, > unset_tz() is supposed to simply remove TZ from the environment. It > does this using putenv("TZ") (SVR4 convention?). However, this has no > effect in FreeBSD (the FreeBSD-ish technique for deleting an > environment variable is unsetenv("TZ")). So when the (usual) case > arises of there being no TZ environment variable when Evolution is > started in FreeBSD, then TZ=UTC ends up being inserted into the > environment whenever certain views are selected and is never deleted > afterward. > > The solution (untested anywhere but on my FreeBSD desktop) seems to be > to change putenv("TZ") to unsetenv("TZ") in... > > libical/src/libical/icaltime.c > > According to man pages, this is valid for Linux as well as FreeBSD > although I notice that early Solaris systems lack unsetenv(). I don't > have a recent vintage Solaris system to check. > > Or an easy workaround would be to simply add TZ to the environment > before running Evolution.
Funny you should mention this. I already have patches, and I will be testing them soon. Look for an Evo update sometime in a few hours. Joe > > > Bill > > On Sun, 2003-08-24 at 22:11, Joe Marcus Clarke wrote: > > On Sun, 2003-08-24 at 18:33, William Bloom wrote: > > > I've just installed the Evolution 1.4.4 port and I'm generally > > pleased, > > > with reservations. I've encountered the inbox permission problem > > > reported in other posts to freebsd-questions, but also I've noticed > > that > > > there seems to be a date/time formatting problem in the message > > summary > > > view that's not been mentioned in earlier posts. For that matter, > > I've > > > seen nothing referencing this problem in the issue list at > > > http://developer.ximian.org/projects/evolution. The symptom is that a > > > message's date/time is shown as expected when the view is first > > > displayed after Evolution startup. But if -any- other view is > > displayed > > > (i.e. Summary), then on return to the message summary the date/times > > all > > > are evidently displayed with a UTC time offset (0) instead of the > > > correct one. Once this occurs, the date/time never returns to the > > > current offset until Evolution is stopped and restarted. Has anyone > > > else seen this as well, or know a workaround? > > > > Ah, that's the catalyst. There have been numerous reports of this > > problem, but I was never able to reproduce it. Switching views is the > > catalyst. I suspect this is not a FreeBSD bug, but now that I have a > > definite catalyst, I will take this to the Ximian people, and see what > > they have to say. > > > > Joe > > > > > > > > Bill -- PGP Key : http://www.marcuscom.com/pgp.asc
Description: This is a digitally signed message part