Frank Ellermann wrote:
On 24 October 2011 16:36, Hector wrote:
seconds = 2DIGIT ; 00-59
totalsecs = 1*(%d48-57) ; total seconds to delay
If this suggestion for a 2nd syntax format is
acceptable, is the ABNF correct?
Ignoring the premise, you can use 1*DIGIT instead of
those decimal constructs ;-) If you want to specify
<totalsecs> details you could suppress leading zeros
and add an upper limit -- the very odd 99-23:59:59
maximum yields 1*7DIGIT if I got this right, please
check it if you want to use it.
I am not expert with ABNF language. Can you tell me what would be is
And trim this to 1*6DIGIT if they let you, IMO 9 days
is already far outside of anything remotely related
I guess we can say the odds are good, it would not be that long or
beyond 4-5 days RFC5321 already recommends and if so, the MTA might
consider to implement an automatic bounce idea after seeing a retry=
hint that goes beyond its local policy maximum days.
So do you think we should?
a) Keep with the proposed retry=[DD-]HH:MM:SS syntax?
b) Provide a 2nd syntax retry=seconds? or
c) Just go with a single, simpler retry=seconds syntax?
With retry=seconds, there would be no technical upper limit, only
policy based for the MTA to determine.