More than 8 consecutive coil addresses results in the coils with index above 8 to not be detected/polled

[Ruaktom]

When I have more than 8 consecutive coil addresses in the confi.yml file, this results in groups being created with more than 8 coils.
The coils with index number above 8 are no longer triggered.

Issue can be solved by introducing an additional condition in the GroupCoils function (coilgroup.go) that creates a new group if size of group would exceed 8.

if coil.Address == groups[groupIndex].offset+uint16(len(groups[groupIndex].coils)) && uint16(len(groups[groupIndex].coils)) < 8 {
	// Add the current input to the current group
	groups[groupIndex].coils = append(groups[groupIndex].coils, coil)
} 
else {
	// Start a new group
	groups = append(groups, CoilGroup{offset: coil.Address, coils: []Coil{coil}})
}

Maybe something to be implemented?

[mhemeryck]

Hi @Ruaktom! Thx for reaching out. I haven't been looking into this repo for quite some time, so apologies for the late reply.

I was just reading into my own source again. Normally, the length of the coilGroups shouldn't be an issue, since they should be mapped back properly to the right individual coil, see the Update function:

Upon closer inspection however I did notice that the modbus client I am using is returning the data as byte whereas I seemed to have assumed it was uint16:

https://github.com/goburrow/modbus/blob/606c02f4eef527a1d4cf7d8733d5fd7ba34f91d8/client.go#L41

Hence, the mapping back I did in the Update function probably does not match:

This should probably fix it:

diff --git a/coilgroup.go b/coilgroup.go
index 67b65c4..2b2d5d3 100644
--- a/coilgroup.go
+++ b/coilgroup.go
@@ -27,8 +27,8 @@ func (coilGroup *CoilGroup) Update() (err error) {
 		return
 	}
 	for k := range coilGroup.coils {
-		numberIndex := k / 16
-		bitOffset := uint16(k % 16)
+		numberIndex := k / 8
+		bitOffset := uint16(k % 8)
 		value := results[numberIndex] & (1 << bitOffset)
 		coilGroup.coils[k].Update(value != 0, coilGroup.MQTTClient)
 	}

I'll see if I can create test to reproduce the issue plus look into a patch release.