Re: [Comment on WS-I18N WD]

Hi Frank,

Frank Ellermann wrote:
> Dan Chiba wrote:
>
>   
>>   2. Timezone (Olson ID or RFC 822 zone offset)
>>     
> [...]
>   
>> Is this a reasonable list of the element items now?
>>     
>
> I'm not sure about "RFC 822 zone offset", that is
> for my European eyes rather US-centric, and not what
> I'd expect in a memo claiming to be about I18N.  The
> draft says:
>
> | Note that RFC 822 zone offsets are not complete
> | time zone identifiers
>   
Yes I think this is an important note. Zone offsets are not very useful 
because it does not identify a time zone. I think this note implies 
using Olson ID is preferred. (for this reason I intentionally put it 
first, switching from the order in the draft).
> OTOH that is unfair, RFC 822 can do any offset down
> to minutes.  But this was later improved in 2822upd
> (soon to be approved) among others:
>
> | zone  =  (FWS ( "+" / "-" ) 4DIGIT) / obs-zone
>
> That is far better, the cruft is in <obs-zone>, and
> you could import 2822upd <zone> excl. <obs-zone>.
>
> The prose is also better, e.g., it specifies -0000,
> and it notes that the last two digits are limited
> to 00..59, for the details compare
>
> http://tools.ietf.org/html/draft-resnick-2822upd-06#section-3.3
>
> There is also RFC 3339 specifying the "Internet
> Date/Time format" (a no-nonsense subset or "profile"
> of ISO 8601:2000).  RFC 3339 specifies:
>
> | time-hour       = 2DIGIT  ; 00-23
> | time-minute     = 2DIGIT  ; 00-59
> [...]
> | time-numoffset  = ("+" / "-") time-hour ":" time-minute
> | time-offset     = "Z" / time-numoffset
>
> You would not want the colon, because RFC (2)822(upd)
> doesn't have it, but you might want the "Z", a part of
> <obs-zone> in 2822upd.  Odd, the IETF USEFOR WG ended
> up with adopting a 2822 <zone> excl. <obs-zone>, but
> keeping one obsolete identifier, in that case it was
> "UT", not "Z" (meaning the same thing, UTC).  The 
> "UT" was for NetNews backward compatibility, nothing
> you need to worry about. 
>   
I quite agree with the idea to exclude <obs-zone>. Allowing "Z" would be 
also good, because normalizing datetime values to UTC is very often a 
good idea.

Regards,
-Dan
>  Frank
>
>
>

Received on Tuesday, 24 June 2008 07:45:00 UTC