Hi Rob.
��
Thank you, Allex, for your comments. When time permits, I will separate the
issues into different posts.
As a first step I have put code for two devices on Github:
- https://github.com/robvanlopik/PCA9685-PWMExpander
- https://github.com/robvanlopik/DS1307-TinyRTC
It would be cool to push it to into PharoThings repo.
��
While working on the PWM module, it occurred to me that PharoThings has two
different points of view. The first I would call "board-centric", where we
tell the board what to do with its pins. That is also the way my
PotPCA9685Device works. On the other hand there is the "pin-centric" view
where pins have functions and are told what to do. This view is used in the
inspectors. As the outputs of the PCA9685 look very much like GPIO pins, I
wondered whether it would be worthwhile to model them as such. This doesn't
seem possible in a direct way, so it would need the introduction of a
PotBoardExpander. Maybe PotBoardConnector could be adapted for this
purpose, but I think it should be something different.
This reasoning also stems from my feeling that the Arduino is not really a
PotBoard. For one thing, there is no reason why it would have to be a
singleton; I can attach Arduinos through numerous serial and USB ports (and
Bluetooth).
The #current variable is just for convenience. The main case it to��have a board instance representing the actual physical thing where the image is running. So we can save the image and the "current board" will react on startup/shutdown��events. Also we can access it from the remote image.��
Nobody should really use this singleton. Any logic should be modelled as devices where you have a board variable. At least that was an idea in my original design.��
In case of arduino board it obviously don't need a current instance as nothing is running there. So it is the responsibility of user to maintain the connected arduino��boards. PharoThings does not provide any model for that (not currently).����
��
On the other hand it would very well fit the description of a
PotBoardExpander. In that case PotBoard should have a superclass
PotController that allows for having no pins of its own.. so any computer
can control one or more Arduinos (To be precise, by "Arduino" I mean a
Arduino-like
Arduino-like board running the Firmata sketch and communicating through the
Firmata protocol).
The reason one would want to do this, is to have a uniform way of
controlling pins, irrespective of board or device. Maybe this uniformity
can also be achieved by implementing the protocol of PotBoardDriver for any
device that has GPIO-like pins.��
Yes. I think it is the right idea.
I think implementing this kind of board expanders as devices is a good solution. But we��need to introduce the notion of board extensions. The board extension is a special board (PotBoardExtension) plus a special driver (PotBoardExtensionDriver). The driver is created on extension device (PCA9685Device) which provides the actual access to extra pins.
This way all logic will be in devices. They implement "extra pin protocol" and provide configured�� board as an extension using reusable classes. Something like:
PCA9685Device>>configureBoardExtension
boardExtension := PotBoardExtension from: self.
boardExtension addNewConnector: 'E1' with: {��
3.3 powerPin. 5 powerPin.
17 gpio. 18 gpio: PotPWMFunction}.
Then we will need to extend board inspector with a knowledge about connected boards provided by devices. So we will be able to inspect them and access pins like in main board.