Re: #16 [*] EPOCH (section 5.1.2)

KAMADA Ken'ichi wrote:
>>#16 [*] EPOCH (section 5.1.2)
>>
>>	Perhaps you need a description of the time stamp?  I recall there
>>	being such a description in draft-housley-binarytime-xx.
>>	(Sam Hartman)
> 
> 
> The current draft specifies it as "the standard posix four octet
> time stamp".  Is describing in KINK preferred to referring POSIX?
> 
> FWIW, the EPOCH field is only used for DPD and the format of it
> is not important, as far as the same value is not reused within a
> short period.
> 

I agree with Kamada-san.

I think it is important for EPOCH to have an implementation determine peer's 
rebooting. DPD is done by detecting the change of a EPOCH in a received message.
Therefore implementation just has to compare a new EPOCHwith old one as a 32bit
binary, so that the format is not important.

Even if we adopt the format which counts up a second from a base time, it is not
useful for a KINK implementation which can reboot in one tenth second.

As Kamda-san said, it is important the value for EPOCH must not reused within a
short period for the purpose.

Anyway I think the formant is not important and it dpends on a system in which a
KINK implementation runs.

I think EPOCH is a value under these conditions.
1. It is 32bit binary.
2. It MUST NOT reused within a short period.

--
Kazunori Miyazawa