Re: Weird timezone handling

[Derek Price]

Stepan Kasal wrote:

>Why the command line is not interpreted correctly in case 2)?
>  
>

Probably because you are seeing the timestamp on the committed file,
which should be the most current as of the time you specified (not
necessarily different for two specified times, especially times only 2
hours apart), and not the time you specified, but I have to run and
don't have time to do the calculations on your data and make sure.

>Why the "rdiff" output uses dates like "Sat Sep 10 08:00:14 2005", which
>is probably in -0400 (the timezone of the server?), while "diff" uses
>always UTC (-0000)?
>  
>

This is a bug, if it still exists in the 1.11.21 server, it should be fixed.

>Why cannot the output use the client's time zone, in all cases?
>(Perhaps including the output of "cvs log".)
>  
>

Because until recently, the server did not include timezone information
with most datetimes, and even the datetimes with tz info were not
flagged as anything other than text for the client to print by the server.

The latest feature releases have infrastructure to do this, but I
believe only `cvs ls' and `cvs log' are making use of it yet.

>Is there a documentation which would explain this dark issue to me?
>  
>

The short form is that with CVS releases earlier than 1.12.10 or so
almost always output time in UTC (though often without printing TZ info)
but clients always interpret times from the command line as from the
local timezone.  Anything else is a bug.  With a 1.12.9 server, some
commands will add timezone info to times output by the server.  With a
1.12.10 client, the client will convert these times into the local zone
for output.

Regards,

Derek

-- 
Derek R. Price
CVS Solutions Architect
Ximbiot <http://ximbiot.com>
v: +1 717.579.6168
f: +1 717.234.3125
<mailto:derek@ximbiot.com>