Re: [core] Designs to resolve streaming issues in SenML

[Carsten Bormann <[e-mail]>]

Cullen Jennings (fluffy) wrote:
> pro / cons of the -03 design vs -04 design.

One comment I made when this was discussed in Yokohama was that the -04
design makes the processor examine every single measurement record
whether there is a change to the base values (and, if there are only
changes to base values so that record isn't really a measurement
record).  That might impede high speed processing a bit.  E.g., it could
become harder to merge two time sequences because each single entry in
them might be changing a base value that is relevant to entries in the
other stream.  But I haven't implemented either model, so that is just a
hunch.

Apart from that kind of consideration, that part of the approach -04 is
pretty much equivalent to -03.

Grüße, Carsten