[Freeassociation-devel] libical fixes for review (and commit if ok please)

[Patrick Ohly]

On Fri, 2009-08-07 at 10:53 -0400, dothebart wrote:
>         
>         http://bugzilla.gnomeorg/show_bug.cgi?id=562028
>         https://sourceforge.net/tracker/index.php?func=detail&aid=2317435&group_id=152976&atid=785987
>         
>         Hurray for fixing it! I assume that it is fixed for more than
>         just
>         Europe/Zurich ;-)
>         
>         devel at lists.sourceforge.nethttps://listssourceforge.net/lists/listinfo/freeassociation-devel
>         
>         
> 
> Patrick, please note that libical.sf.net is a different project, so
> you're probably barking at the wrong tree with your bugreport...

Yes, I must have picked the wrong project at that time. However, it
definitely is in the system timezone to VTIMEZONE conversion code
written by the Evolution devs, so it ended up in the current libical. 

> Is that bug covered by any of the already available regression tests?

No, I don't think so. The timezone test program that I wrote only checks
that the day of the transition is correct in the generated VTIMEZONE,
but it does not check for the exact hour. That's where the bug might
have been found.

>  since regression.c for example says
> 
> not ok 24 - America/Los_Angeles local time is 2000-11-03 10:30:30
> # test failed: "(strncmp(ictt_as_string(icttla), "2000-11-03
> 10:30:30", 19)==0)"
> #          at: regression.c:1627

I think the tests are outdated. I haven't checked it again, but remember
that I saw the same failure a while ago.

The tests rely on having the correct time zone rules for 2000, but the
rules have changed and the generated VTIMEZONE is only guaranteed to
work correctly for next twelve months:
http://bugzilla.gnome.org/show_bug.cgi?id=540676


-- 
Bye, Patrick Ohly
--  
Patrick.Ohly at gmx.de
http://www.estamos.de/