info-cvs
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: cvs up slowdown


From: James Cloos
Subject: Re: cvs up slowdown
Date: Thu, 21 Sep 2006 06:39:30 -0400
User-agent: Gnus/5.110006 (No Gnus v0.6) Emacs/23.0.0 (gnu/linux)

>>>>> "Mark" == Mark D Baushke <address@hidden> writes:

Mark> It does not sound to me as though posix/UTC is useful if it is
Mark> different than the TAI (Atomic Time) derived from cesium clocks.

Agreed.  None the less it is what one gets by default on posix boxen,
unless a right/ zonefile is specified.

Mark> The current difference between TAI and UTC is 33 leap seconds
Mark> as of the 2006. (see http://www.ece.udel.edu/~mills/leap.html).

Mark> If posix/UTC time is only 23 seconds from UTC, then where are
Mark> the other 10 seconds?

POSIX time matches UTC at 1970 Jan 1.  The other 10 leap seconds
predate 1970.  GPS time does the same thing:  it matches UTC at
its epoch and diverges as of the next leap second after that.

So, in fact, TAI, GPS, POSIX and UTC are all different times.

Mark> The CVS sources try to do their best to ignore the local
Mark> timzeone.

Of course.  But I need to think a bot more about the rest of these
details before replying to them.

Mark> For some reason, my GNU/Linux box sees a
Mark> difference of 22 seconds instead of the 23 seconds you see.

You probably have an older version of the zonefiles.  The last leap
second was 2005 Dec 31 23:59:60 UTC; your zonefiles probably predate
the announcement of that leap second.  (Ie they are at least fifteen
months out of date. :)  (Many dists have now split the zonefiles into
their own package so that they can be updated more regularly than libc
is generally updated.  That started happening last year after the
announcement of the 2005 leap second.)

One thing I just realized, though, is that the problem only occurs if
there is a change between posix and right timezones since the last up
or co.  Ie, if the co (or the last up) is done under a posix tz and
then up is run under a right tz, or visa versa.  The repos I first
noticed doing this were created by ebuilds which grab the source
from cvs rather than a tar.  Obviously gentoo's emerge must unset
or override TZ.

I'll do some more testing and get back.

-JimC
-- 
James Cloos <address@hidden>         OpenPGP: 0xED7DAEA6




reply via email to

[Prev in Thread] Current Thread [Next in Thread]