[Ntp] Antw: [EXT] comments on draft‑mlichvar‑ntp‑ntpv5‑03

>>> Dan Drown <dan-ntp@drown.org> schrieb am 23.11.2021 um 20:15 in Nachricht
<20211123131501.Horde.ErUH7VWw3Nr2PFkAGzGIEuI@mail.drown.org>:
> After reading this draft, I'm glad to see these things are included:
> 
> 1. the unknown leap seconds flag seems useful for embedded systems  
> that don't have easy access to that information
> 2. the explicit interleaved mode with server cookies
> 3. client timestamps are switched to a client cookie
> 4. the removal of all modes except client/server
> 
> I have a few comments, and I think I'll break them up into multiple  
> emails to make discussion easier.
> 
> This email will be on the "3. Data Types" section.
> 
> For the time16 definition "the maximum value is 32767/32768" This  
> confused me on first read because I parsed it as either/or at first  
> which made no sense. I had to think about it until I realized this was  
> talking about a fraction. I'm proposing a change to: "the maximum  
> value is the fraction 32767/32768 (almost 1 second)"
> 
> Should the log2 format be explained in more detail in this section?  
> I've seen a lot of people confused about this data type, so I'm hoping  
> a definition would help. Maybe something like "log2 ‑ an 8 bit signed  
> integer representing the rounded log2 value of seconds. For example, a  
> log2 value of ‑2 would be 2 to the power of ‑2 or 0.25 seconds"
> 
> Should the change from NTPv4's time32 format be mentioned here?   
> Something like:
> "A 32‑bit fixed‑point type containing values in seconds. It has 4  
> unsigned integer bits and 28 fractional bits. The maximum value is 16  
> seconds and the resolution is about 3.7 nanoseconds. Note: this is  
> different than the 32 bit time format in NTPv4 (RFC5905)"

I wonder: Up to NTPv3 floating-point math was avoided, but starting in NTPv4
floating-point math is used.
Do we want to use floating-point math in NTPv5, too? What about those embedded
systems?
If so, are those many special fixed-point formats justified (outside of
packets)?

Regards,
Ulrich

> 
> _______________________________________________
> ntp mailing list
> ntp@ietf.org 
> https://www.ietf.org/mailman/listinfo/ntp 