[6486] in Kerberos
Re: More gmt_mktime trouble: breaks on systems with leap seconds
daemon@ATHENA.MIT.EDU (Jonathan Kamens)
Wed Jan 17 00:44:01 1996
Date: Wed, 17 Jan 1996 05:37:23 GMT
From: Jonathan Kamens <jik@annex-1-slip-jik.cam.ov.com>
To: jhawk@MIT.EDU
Cc: kerberos@MIT.EDU
In-Reply-To: <199601170044.TAA02459@zygorthian-space-raiders.MIT.EDU> (message
from John Hawkinson on Tue, 16 Jan 1996 19:44:52 -0500)
Date: Tue, 16 Jan 1996 19:44:52 -0500
From: John Hawkinson <jhawk@mit.edu>
It would be a bad idea for Kerberos to have kludgey leap second
support, though, particularly since they aren't POSIX-compliant.
It makes no difference to me whether gmt_mktime() uses the code I
added or posix2time() to solve the problem I reported. If I'd known
about posix2time(), I would have used it instead.
(And perhaps the POSIX people should get with the program and
acknowledge that leap seconds exist and update their standards. :-)
I suppose I ought to include the obligatory flame that the kerberos
team (of which I'm not a member) frowns on #ifdefs that aren't
autoconf derived, and submitting patches containing strings like
"#ifndef PUNT_LEAPSECONDS" is likely to make them less than pleased.
To be blunt, big whoop. It'll take them a whole five seconds to
remove those lines, and I spent a bit longer than that working on the
fix, and I in essence donated my time to them by submitting the fix to
them.
I do try to conform to the conventions that the maintainers of a
package maintain when submitting packages to them. However, in this
case, I think the difference between what I sent and their conventions
is minimal and easily rectified, and it would have taken me a lot more
time to come up with configure changes to do the appropriate thing,
and as you pointed out, it would have been wasted effort anyway since
I should have used posix2time().
Jonathan Kamens | OpenVision Technologies, Inc. | jik@cam.ov.com
