Tim Goodwin | 1 Mar 1995 14:49
Picon
Favicon

Re: Internet draft draft-onions-822-mailproblems-00.txt

[ This message arrived on the smtpext WG list.  I think mailext is
probably the right group, so am copying there.  Please send replies to
mailext <at> cs.wisc.edu (I haven't set `Reply-To:', as that seems to upset
the Wisconsin listserv). ]

> from scanning the date: fields for message sent to me over the past
> 5 years (that I have kept)
> 
> 9496 - RFC compliant

I'm extremely sceptical of this figure.

Note that RFC-822 defines only 10 valid alphabetic time zones (plus the
military ones, which were deprecated by RFC-1123, although I still see
them occasionally).  `Date:' fields containing timezones like `BST' are
*not* RFC compliant.

>  675 - no seconds

Seconds are optional.

     hour        =  2DIGIT ":" 2DIGIT [":" 2DIGIT]

>  462 - no Day

The day is optional.

     date-time   =  [ day "," ] date time        ; dd mm yy

>   60 - no Day, no seconds, no timezone
>   10 - just DayNum Month Year            eg 28 February 1994
>    5 - no seconds, no timezone
>   17 - randomly mis-formatted 17 different ways !

> Another common problem with numeric timezones is some UA's put in the 
> offset like one would write it in the 24-hour clock (+100), whereas others 
> assumed that you could only have whole-hour offests (as +1).

Neither of these conform to RFC-822, which clearly states that a numeric
timezone must be 4DIGIT.  Why do people find it so hard to implement
this correctly?

The "best" malformed `Date:' fields I've seen recently claim a timezone
of `+-100'.  This appears to be generated by Microsoft's so-called SMTP
gateway.

Tim.


Gmane