Hi Chris,

thanks for bringing that up.
I think it’s a bit similar than our "pseudo symbolic addressing" idea for (old) 
S7 Driver.
So I am strongly for an additional (query-)parameter for the filenames.

Julian

Am 22.11.19, 14:20 schrieb "Christofer Dutz" <[email protected]>:

    Hi,
    
    while working on the KNX and the BACnet support I did come across, that 
usually you only get sensible and usable data, if we enrich this with data from 
some other source.
    For BACnet this is a so-called EDE file and for KNX I had to add a parser 
for the ETS5 knxproj files.
    
    For KNX for example the protocol provides raw byte data which can only 
sensibly be decoded if you know the type. You can only get the type from the 
group-address a piece of data is sent to.
    How to decode the group address is also controlled externally in thr 
knxproj file. So if that says, this house is using 3-level group addresses, I 
know how to decode the two bytes of the destination address.
    Then I can use this to lookup the type and can use that information to 
decode the payload.
    
    In SteamPipes I used a so-called Processor to enrich the raw data-stream. 
But when using this sort of thing in a PLC4X driver, how do we do this?
    
    Should I add multiple optional parameters:
    
      *   knxproj-file: to get the full enrichment
      *   group-address-encoding-level (1 | 2 | 3): to at least be able to 
decode the destination addresses
    
    I think it doesn’t make sense to generally add a layer of abstraction to 
these drivers that sort of wrap the driver … what do you think?
    
    Chris
    

Reply via email to