St?phane See below for my response.
St?phane VOLTZ wrote: > Hello, > > I have borrowed a Dell A920 which happens to be a relabelled Lexmark > X1100 >series scanner. As expected, it is slightly different from the X1100 model >for which the lexmark backend works. So I'm about to add support for the A920 >in this backend. For that I think doing the following changes: > > - move the shadow_registers table in device struct in case someone has > several of theses scanners plugged int the same computer. > > Yes. This makes sense as I've moved just about everything else under the device struct for the same reason. > - detect scanner model by using USB ids > > Sounds good. I assume the Dell A920 has a different ID than 0x043d 0x007c and that there's an example of how to do this in other backends. > - add a struct holding flags and parameters on a model basis like the > gt68xx > or genesy backend > > I haven't looked at these backends. I suppose this would be okay. What flags and parameters would differ between models? > - move constant register values to sanei_lexmark_x1100_init(), and > initialize > differently according to scanner model > > Hmmmm. This is something I should have probably done earlier as I think I still set the constant register values per scan operation. Care will have to be taken that nothing breaks when moving this to sanei_lexmark_x1100_init(). > - finally set up the about 20 registers (sensor and motor related) that > differs between the X110 and A920 in each scan opertion > > I assume this will be handled conditionally (if x1100 do this..., else if A920 do this...) in the appropriate function for each scan operation. > Fred Odendaal, what do you think of it.? If it is the correct route to > take, >how do you prefer these changes being done ? Through patches sent to the >list, or in a copy of the backend in the experimental CVS staging area ? > > I'd rather do it through the experimental CVS staging area. Do we need to do a branch? regards, Fred O. >Regards, > Stef > > >