Function pointers

a_riot42

One thing that Netlinx is missing that I miss from C is function pointers. Has anyone tried to implement something similar? I could use it in this job where I have 100 functions that will activate any one of 10 devices in 10 zones. So for device 1 which is a DVD I have 10 functions, one for each of the 10 zones. ie: kitchenDVD, mBedDVD, livingDVD etc.

However, if I want to start the DVD in all 10 rooms at once, I have to write a switch that takes the zone and then calls the appropriate function for each zone.

if (device == DVD)
{
  switch(output)
  {
    case mBed:
    {
      mBedDVD()
    }
    case kitchen:
    {
      kitchenDVD()
    }
    ...
    ...
  } 
}

if (device == DVR)
{
  switch(output)
  {
    case mBed:
    {
      mBedDVR()
    }
    case kitchen:
    {
      kitchenDVR()
    }
    ...
    ...
  }
}

Its alot of code to do something pretty basic.

What I would do in C, is have an array of 100 function pointers so that for each input/output combination I just call the function that the pointer in the array points to.

for (int i = 1;i <= 10;i++)
{
  call functionptrArray[i]    // approx in Netlinx
}

I can't figure out a way to do this in Netlinx because there are no pointers but I am wondering if someone has implemented something similar with virtuals, events etc?

I won't know at compile time what device/zones the user will choose, so I need a way at runtime to change the function that is called depending on the device and zone. I was hoping that there would be a better way than hard code all the function calls.
Thanks,
Paul

mpullin

Whereas I too miss the presence of pointers in NetLinx, but I don't think you need them to improve your code example. Try making a single function that takes two integer parameters, one for the device and one for the zone, e.g.

DEFINE_FUNCTION TURN_ON(INTEGER zone, INTEGER src){
     switch(src){
          case 1: // dvd player
               switch(zone){
                    case 1: // bedroom?
                         // turn on bedroom DVD
                    break;
               }
          break;
          case 2: // etc.
          break;
     }
}

Now you can create loops to make calls to that function if you wish.

a_riot42

That was what I was trying to avoid. Doing it that way requires 100 different cases, which isn't too bad I suppose, but if I had 12 devices and 20 zones then I am now at 240 cases. Since every zone/device function requires turning different devices on or off I need 100 different functions to do every possible combination. I thought there might be a more compact way of doing this. Here is how I turn on say 10 zones to the DVD.

/* pseudocode */

button_event[dvTP, 100]
{
  for (cnNumZones)
  {
    currentZone = currentZone + 1
    do_push(dvFamilyTP, 20)
  }
}

Right now I have it set up where my button events for devices contain a switch like so:

button_event[dvTP, 20]
{
  push:
  {
    switch(currentZone)
    {
      case cnMBed:
      {
        mBedDVD()
      }
      case cnKitchen:
      {
        kitchenDVD()
      }
      ...
      ...
   }
  }
}

Is it possible for the timing of the do_push and the increment of the currentZone variable to get screwed up such that the for loop finishes before the do_push gets executed? I assume button_events are executed on an as needed basis so you have no way of knowing exactly when (it terms of where the code is) it starts and finishes.
Thanks,
Paul

Colzie

I create structures. u_zone, u_display, u_source, etc., and initialize these upon startup with all the info I need (device, name, etc.). Then I can create functions (one for displays, one for dvds, etc., each taking parameters such as device (or device pointer?) as well as the command) to do things like

send_string u_display[iDisplay].dv_device, iCmd
or
pulse[u_source[iSource].dv_device, iCmd].

Doing it this way can eliminate a lot of switch/cases.

a_riot42

Colzie wrote:

I create structures. u_zone, u_display, u_source, etc., and initialize these upon startup with all the info I need (device, name, etc.). Then I can create functions (one for displays, one for dvds, etc., each taking parameters such as device (or device pointer?) as well as the command) to do things like

send_string u_display[iDisplay].dv_device, iCmd
or
pulse[u_source[iSource].dv_device, iCmd].

Doing it this way can eliminate a lot of switch/cases.

Thanks, I do something similar, but the issue is that I want to call a function not send a string, command or pulse. I am guessing there is no way to do this other than come up with some other scheme. There doesn't seem to be a way to dynamically call different functions at runtime without hard coding it all.
Paul