On 22 Oct 2019, at 10:50, Carsten Bormann wrote:
On Oct 22, 2019, at 17:33, Pete Resnick <resnick@xxxxxxxxxxxx> wrote:
Implementations are perfectly capable of multiplying and adding.
I think after 40+ years in the industry I can refute that statement
:-)
The can successfully write code to convert to SenML, but can't write
code to do simple arithmetic? Color me unconvinced. ;-)
But, seriously, this draft is based on a specific request from an SDO,
OMA/IPSO, for which this secondary registry would enable porting their
existing data models to ones based on the SenML units registry. Other
SDOs from this space might follow suit.
Clearly, when we did SenML, we didn’t want to go this route.
Currently, there is an amazing pull towards data model harmonization
that has changed the fundamentals enough that this draft now makes
sense. I’ll leave it to IPSO people to explain this in more
details.
But if that's true, it's also likely that these units will leak through
to base RFC 8428 implementations. Do not make it optional for those
implementations to deal with the new units; add them to the existing
registry.
“MAY, but SHOULD NOT” means exactly what it says; I don’t think
there is a contradiction.
But of course we can contract this to “SHOULD NOT” if that helps.
Save yourself the time of having Barry explain the problem to you in
excruciating detail during IESG Review. :-)
That said, this is obviously not the path I want you to take.
Oh, and “cm” isn’t registered yet because OMA didn’t need it,
not because it couldn’t be registered if that turns out to be
desirable.
I would put pretty good odds on it appearing pretty soon after this
document is released; implementers will assume that any unit will do.
SenML is now in a position to play a centerpiece in the harmonized IoT
data landscape. Not going for this draft is likely to thwart this
opportunity.
Sounds like a fine thing to aim for. Again, that says to me to add these
to the existing registry.
pr
--
Pete Resnick http://www.episteme.net/
All connections to the world are tenuous at best
