In the end I think the generated drivers are now easier to use. If an argument is required you now can see it in the signature and don't have to pass in an Object[] and hope to put the right object type in the right position. The root elements should never need any additional arguments.
Chris Am 24.09.19, 11:22 schrieb "Julian Feinauer" <[email protected]>: Looks good to me, although I'm not able to follow as closely as I'd like to do : / Julian Am 24.09.19, 11:19 schrieb "Christofer Dutz" <[email protected]>: Hi all, I just committed some changes to the code-generation. Here a little summary: * First of all the IO modules are no longer real objects … the parse and serialize methods are now static. * The reason for this was that otherwise I ran into stack-overflow-errors as soon as there are types embedded in themselves (Like in BACnet) * Also it simplifies the code a lot * I noticed there’s actually no need to support an Interface in this case * I added a new PcapChannelFactory which can be used as alternate input for a RawSocketDriverBase * This allows to provide the input to a passive mode driver by providing a PCAP capture file. * This way all you need to develop a passive mode driver is a WireShark capture of the traffic and you no longer need direct physical access to it * You can speed up and slow down the replay I am now thinking of adding a new field type “enum” as currently there is no way to use the Enum types in a language spec. So hopefully the “enum” field will make this possible. Chris
