Add description of "gpio"-tag

[destogl]

No description provided.

[gavanderhoorn]

GPIO as an acronym has a fairly well established meaning. See wikipedia/GPIO for instance.

Is the intention to use it like that in ros2_control, or is it even more general?

The text is slightly ambiguous, as in some places it states that the gpio tag should/could be used for anything which is not a joint or a sensor, but in other places (like here) it limits it to the more traditional meaning of the term (ie: "an uncommitted digital signal pin on an integrated circuit or electronic circuit board which may be used as an input or output, or both, and is controllable by the user at runtime" (Wikipedia)).

[destogl]

@gavanderhoorn This is a very good point.

For clarification. We would like to use it for everything which has not "a physical" signal. The usual use-case would be digital and analog input outputs on a robot that can not be related to any joint or sensors.

Is the intention to use it like that in ros2_control, or is it even more general?

The use is more general because it also describes analog signals.

[gavanderhoorn]

We would like to use it for everything which has not "a physical" signal.

That seems to contradict what's in the document right now (emphasis mine):

Although strictly used for digital signals, in this document describes any electrical analog, digital signal, or physical value.

Perhaps you mean something else?

The usual use-case would be digital and analog input outputs on a robot that can not be related to any joint or sensors.

well, anything which you connect to a digital computer in the end is represented by digital input outputs. So this doesn't seem to say very much.

I understand why you went with gpio, but to me, it it's either not specific enough, or it's too generic.

And the given rationale and use-case seem a bit narrow as well: why only robots? What about using ros2_control for something which is not (directly) considered a robot?

[gavanderhoorn]

@destogl: what about type and width of gpio elements? Is that all left to the implementation "behind" the interface?

[destogl]

@destogl: what about type and width of gpio elements? Is that all left to the implementation "behind" the interface?

We should probably add type. This discussion is related to #39.

Regarding width, I am not sure what would be the use for it. It suggests that we would have to start working with vectors in the backed in the real-time related loop. I think we will keep this on hold until there is a specific use-case for this.

[gavanderhoorn]

Regarding width, I am not sure what would be the use for it. It suggests that we would have to start working with vectors in the backed in the real-time related loop. I think we will keep this on hold until there is a specific use-case for this.

if you are going to use the name gpio, I believe width should be one of the properties. Or you'd have to make it really clear your gpio is not what's typically understood to be gpio and/or clarify that each gpio is always a 1D/scalar value/pin/signal.

Perhaps signal could be an alternative name? It would still make me ask "what's the resolution?" (which in digital contexts is almost the same as width), but at least we'd avoid the use of gpio.

It suggests that we would have to start working with vectors in the backed in the real-time related loop

that's not really a problem.

I would not dismiss design options because they might pose a challenge in the implementation.