[squeak-dev] Time zone help

David T. Lewis lewis at mail.msen.com
Sat May 30 18:52:59 UTC 2009


On Sat, May 30, 2009 at 11:48:14AM -0400, C. David Shaffer wrote:
> David T. Lewis wrote:
> >
> >Hi David,
> >
> >For reasons that I do not understand, the New_York zoneinfo file from
> >your Gentoo distribution is larger than I expected, and contains additional
> >binary data following the time zone table data. My file loader checks
> >for this, and throws up the warning that you are seeing. I don't know
> >if this is some Gentoo-specific thing, or if there is some alternate
> >file format that I am not aware of.
> >
> >  
> Thanks for the patch.  It is definitely not Gentoo-specific.  To make 
> sure I downloaded the Olson database (Gentoo uses tzdata-2009c currenly, 
> although I see newer versions).  I downloaded tzcode-2009b, compiled the 
> code, used it to build a zoneinfo database and the New_York file 
> produced from this is exactly the same as the one I have in my system.  
> What version of zic are you using?  Which data set?

Right you are. It looks like there is now a new version of the tzfile format.
I'm using a SuSE Linux system and a FreeBSD system, neither of which has
this new version. According to the new tzfile man page, the new file format
is backward compatible, and the patch that I sent to you should be safe
to use.

I'll look into this more carefully when I get some time, and update the
TimeZoneDatabase package as needed. Note, I wrote this back in 1999, and
the tzfile parser has not needed any attention until now. Thanks for
catching this and letting me know about the issue.

> With your patch and the attached patch which I had to add in order to 
> ignore some time zones that caused other problems.  I can't recall which 
> ones they were but I know I don't need them :-)

Thanks, I'll look into that also.

> After that I did a:
>    TimeZoneDatabase buildSystemDatabaseForSqueak
> 
> Running the tests gives 14 failures most with names that hint at DST 
> transition issues (testDateAndTimeAtDstTransition, 
> testDateAndTimeImmediatelyAfter...).  Attached is a stack dump for one 
> of the failed tests in case it provides a clue.

I'm seeing the same thing on my system. I believe that the errors are in
my test code, which looks for DST transitions at certain times and dates.
The Michigan legislature changed the law about when daylight savings time
is in effect here, so my tests may be out of date. I'll check this also
when I get time.

Thanks for your help and feedback,
Dave




More information about the Squeak-dev mailing list