On Wed, Jan 02, 2008 at 11:19:42PM +0100, Toni Mueller wrote:
> I'll talk to upstream about it, but personally think that while you
> are formally correct, legibility is actually increased by using a
> dot.
Today I ran into a specific use case where the current behaviour is
clearly detrimental. To wit: GNU date's -d switch doesn't understand
a period separator.
$ date -R -d 2008-07-05.09:17:19
date: invalid date `2008-07-05.09:17:19'
$ date -R -d 2008-07-05T09:17:19
Sat, 05 Jul 2008 12:17:19 +1000
This wasted time when I was trying to RFC 2822ify dates from Darcs'
roundup instance, prior to injecting them (along with other data) into
a mail system.
Perhaps upstream would prefer to adopt RFC 3339 rather than ISO 8601?
It's a significantly simpler standard, and appears to use whitespace
rather than a period or a T (which personally I think is the most
"readable" of the three).
$ date --rfc-3339 seconds -d 2008-07-05T09:17:19
2008-07-05 12:17:19+10:00
GNU date's -d switch also understands RFC 3339 format:
$ date -Iseconds -d '2008-07-05 12:17:19'
2008-07-05T12:17:19+1000
PS: I also notice that the timezone information is missing. If the
BTS is accessible internationally (as is the case for bugs.darcs.net),
this can result in the timestamp being wrong by a whole day!
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
