[ sending to netbsd-bugs list again ]

In article <18220.911205554@munnari.OZ.AU>
kre@munnari.OZ.AU writes:

>   |  1. Rewrite util.c::remotemodtime() using timegm(3) instead of mktime(3).
> Pick this one.   ...
> Or ...
>   |  3. Calculate time_t value by hand.

O.K.  These are what I originally indented to do.
I talked with other developers and I've got no strong objection
(and no good solution seems to be available...),
then I'm going make changes a. to use timegm() on NetBSD
and b. to supply alternative code for portability
(not completely portable, though).


By the way,

> Pick this one.   I don't see that it matters that NetBSD's ftp client won't
> compile trivially on brain damaged systems that don't have timegm(), at least
> until a blessed replacement is defined for that fundamental piece of the
> basic time management software.

It's a lossage of standards, not of system,
and don't say bad about other OSs. :-)
I hope future standards will specify good interfaces
to convert any timezone time to internal time_t value.

Thanks,
--
ITOH, Yasufumi <yasufu-i@is.aist-nara.ac.jp>, <itohy@netbsd.org>