Testing needed for the new date routines

[Russ Allbery]

CURRENT snapshots have the new date generation routines but they're not
being used to generate Date headers yet because I want to make sure they
produce valid results.  So far, I've only tested the code on Solaris.

I'd love for someone to test the code on a *BSD platform (anything that
has tm_gmtoff and tm_zone in struct tm).  convdate with the -d flag will
print out what should be a valid Usenet Date header.  Please follow up to
the list if you can test on a *BSD and let us know if the results look
correct.

On Solaris:

headwall:~> convdate
Thu Jun  8 01:22:52 2000
headwall:~> convdate -d
Thu, 8 Jun 2000 08:22:54 +0000 (UTC)
headwall:~> convdate -dl
Thu, 8 Jun 2000 01:22:56 -0700 (PDT)
headwall:~> env TZ=Iran ./convdate -dl
Thu, 8 Jun 2000 11:55:47 +0330 (IST)
headwall:~> ./convdate "`env TZ=Iran ./convdate -dl`"
Thu Jun  8 01:27:22 2000

I've just checked in the modification to convdate that causes it to assume
the current time if no dates are given on the command line, so it should
be in tonight's snapshot.  The above commands are a reasonable basic test.

(Note that Iran is a good test because it's a half-hour time zone and a
somewhat unusual one; there are several others.  The last test is useful
because it generates a Date header and then feeds it into parsedate to
make sure that it parses correctly.)

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>