Hi Chris, Currently I prefer using on RequestItem for one coil value as it would get too confusing writing coils. For example writing then a 42 to a coil address in modbus would mean setting the 3, 5 and 7th coil. Writing coils requires a quantity which is mapped by the number of items time being. How would you map this in the 42 case? 8 coil fixed? 7 coils depending on the last ‚true‘ value. So IMHO the current solution is the best to just use on item on coil. One thing we could do is something like c logic where 0 =! true. All in all the modbus implementation is still a WIP and maybe we find better solutions when we have more and more practical use cases.
Sebastian > Am 09.07.2018 um 11:04 schrieb Christofer Dutz <christofer.d...@c-ware.de>: > > Ping? Any opinions on this topic? > > > > Chris > > > > Am 05.07.18, 11:15 schrieb "Christofer Dutz" <christofer.d...@c-ware.de>: > > > > Hi all, > > > > Sebastian and I just had a little off-list discussion and I’m taking this > back to the list to get your opinion. > > > > The issue we were discussing is that for example in the S7 driver, we are > mapping all supported PLC4X datatypes to corresponding S7 datatypes. > > > > Now there are generally two options: > > > > * For every driver, create code to map all PLC4X supported datatypes > to corresponding datatypes > > * For every driver to use only the supported types and to throw > exceptions for all unsupported ones. > > > > Now I definitely generally favor the first option. As it’s one of PLC4X’s > unique selling points that you can write code for any PLC without knowing the > details of the protocol used. > > This would be violated by the second option, in my opinion, because > changing from S7 to Modbus for example would require a change of code, if we > were reading a Byte for accessing the 8 Boolean digital input values at once. > > > > If a user wants to for example write a Byte, Integer, Double to a Modbus > Coil (Boolean), then I think PLC4X should map a single byte write request > into 8 Boolean ones (The larger ones mapped accordingly). > > > > Regarding Floating-Point numbers and Time values however I would not 100% > insist on this, because not all PLCs support these and converting a Double > into a Boolean pattern doesn’t really make sense, same with converting a Time > into a Boolean value (Converting this into an Long however could make sense) > > > > What do you others think? > > > > Chris > > > >
