pybricks
[Feature] Hub display dialog / bluetooth advertisement on hub and connected PUP devices status
Is your feature request related to a problem? Please describe. When connecting and experimenting with a hub, the current IDE does not help the students to sufficiently explore their hub and the connected perioferials.
Describe the solution you'd like I would love to have a status dialog displaying the hub status and connected periferials. As a plus I could imagine a "heart program" simple control mechanism or even an EV3 "port view"
Describe alternatives you've considered The dialog I would welcome could display
- connected hub, name, battery, BT signal strength
- connected devices and their values (sensor / motor)
- maybe SPIKE hub led matrix
This would require a constant BT advertisement, that is currently containing only a few values on the hub, nothing on the PUP devices. Similar one is available by the default LEGO implementation over the (tbc) 0000fd02-0000-1000-8000-00805f9b34fb service/characteristic.
Additional context
- EV3 port view: https://ev3lessons.com/en/TabletLessons/beginner/PortView.pdf
my example (IDE only) mock
- https://github.com/afarago/pybricks-code/tree/experiment-hubcentral
I would love to have a status dialog displaying the hub status and connected periferials. As a plus I could imagine a "heart program" simple control mechanism or even an EV3 "port view"
Fully agreed! :+1:
UX thoughts This mainly boils down to whether we want to have Port View while the program is running too, or just when idle.
Allowing it only while no program is running is probably the simplest, as discussed below. This is also the best for the user, I think. Live during a program introduces unwanted lag.
UI ideas
- Instead of adding a new button at the top, I think we could show your suggested dialog when the user clicks on the connected hub at the bottom right. The small status view we have currently could go away.
- Instead of abstract icons, we could also consider showing something a bit like this. In particular, it would be nice to intuitively visualize the 0 point of the motor instead of showing just a degree value.
Implementation idea 1: Builtin port view program There is potentially a comparatively easy way of doing this without inventing new protocols. Opening the dialog could trigger a builtin program on the hub to run, just like triggering the REPL. Closing the dialog or pressing the hub button stops it.
That builtin program could be implemented entirely in Python. It would send all the sensor values to the PC. Somewhat similar to stdout, though perhaps using https://github.com/pybricks/pybricks-micropython/pull/246 so the user doesn't see it in the terminal output window.
These sensor values could be formatted strings like -379° that can be displayed as-is, avoiding the need for protocols or logic that could break between updates on the firmware or IDE side. Or perhaps "A,132,-379°" to indicate the port, device/icon ID and the value to display.
Likewise, https://github.com/pybricks/pybricks-micropython/pull/246 can also be used the other way to select a sensor mode if we want to support that. For the color sensor specifically, we can already show reflection and color and a live colored box representing the HSV value without changing any modes. Modes would just be needed for ambient sensing.
And since this approach uses a builtin program, it can also work offline to display values on the hub display like the heart program you suggested. Maybe a particular button combination (left + right) could start it, or we could expand the the hub UI a bit.
First of all, thanks for the prompt ideation, you are awesome.
Let's focus first on getting the data out of the hub, then we can focus on the UX aspects. Agreed - it makes much more sense to only make this data available when the hub is connected and idle. This renders any custom BLE unnecessary - agreed. It might not make sense to have this always available (LEGO SPIKE App always shows sensors even in the code editor window) - I do not think this is useful.
Following up on this having a piece code available on the brick sending the data is a valid concept.
As I understand we can evaluate two alternatives
- have a port view sender module part of the firmware and connect to this
- when starting the port view window a custom code is downloaded via ~REPL
Any preferences? By myself I think I can progress by creating this python code to report this data. Can I get some pointers how I could accomplish from the react app a REPL start and download (or any alternatives)?
when starting the port view window a custom code is downloaded via ~REPL
Just to clarify, I meant that the program can be a Python module that we include in the firmware, so there isn't any download time. What I mean with "start like the REPL", is that we could have a similar command to starting this builtin Port View program. You can have a look around Pybricks Code to see how the REPL is currently activated. (Unlike most MicroPython variants where the REPL is something special, a bit like the OS, we just treat the REPL like any other user program. It just so happens to use stdin/out a lot, but otherwise isn't special.)
We could start by just writing and starting it like normal Python code. We could also start by printing the result to standard out, and start thinking about merging https://github.com/pybricks/pybricks-micropython/pull/246
I am coming up with the hub side program. Is this something that we could think about? I am not sure about the firmware constraints, therefore I was rather thinking on the REPL on demand download.
This is the sample output that I would channel to the pybricks-code IDE.
H,Luna,3.4.0; Pybricks MicroPython v1.20.0-23-g6c633a8dd on 2024-04-11,SPIKE Prime Hub with STM32F413VG,7430
H:BAT,8393,114,0
H:BUT,
H:IMU,0,1
P,A=63,B=61,C=48,D=x,E=63,F=x
P:A,63,0
P:B,61,-1
P:E,63,0
P:E,63,1
P:E,63,48
P:E,63,53
P:E,63,1
P:E,63,0
H:BUT,B
H:IMU,-1,0
P,A=63,B=x,C=48,D=x,E=63,F=x
H:IMU,-1,1
H:IMU,-1,0
H:BUT,C
Any hints how I can channel the terminal output to the dialog? I will experiment around React, yet any hints are more than welcome.
This is the code: https://gist.github.com/afarago/350a931992e05c423860f4b25ea5c6a2
Hi Attila,
Not really in line with your question, but pylint found two syntax errors in your program: Both times quotes in an f-string. On line 54:
StatusReporter((lambda : f"H:BUT,{"+".join(sorted(str(b).replace("Button.","")[0] for b in hub.buttons.pressed()))}"), None, 100),
Might be (6 double quotes replaced by single ones):
StatusReporter((lambda : f"H:BUT,{'+'.join(sorted(str(b).replace('Button.','')[0] for b in hub.buttons.pressed()))}"), None, 100),
And similar on line 80:
retval.append(f"{portid}={self.deviceids[idx] if self.deviceids[idx] else "x"}")
Might be:
retval.append(f"{portid}={self.deviceids[idx] if self.deviceids[idx] else 'x'}")
I also ran your program on Technic hub and city hub. Works OK.
Interesting program, thank you,
Bert
Hi Attila,
Not really in line with your question, but pylint found two syntax errors in your program: Both times quotes in an f-string. On line 54:
StatusReporter((lambda : f"H:BUT,{"+".join(sorted(str(b).replace("Button.","")[0] for b in hub.buttons.pressed()))}"), None, 100),Might be (6 double quotes replaced by single ones):
StatusReporter((lambda : f"H:BUT,{'+'.join(sorted(str(b).replace('Button.','')[0] for b in hub.buttons.pressed()))}"), None, 100),And similar on line 80:
retval.append(f"{portid}={self.deviceids[idx] if self.deviceids[idx] else "x"}")Might be:
retval.append(f"{portid}={self.deviceids[idx] if self.deviceids[idx] else 'x'}")I also ran your program on Technic hub and city hub. Works OK.
Interesting program, thank you,
Bert
Thanks! Stupid question - how do you toolchain pylint? I am using the code.pybricks.com now, that is awesome, yet pylint and auto format is not av.
Any hints how I can channel the terminal output to the dialog?
We will either need a new event type or status bit to indicate that this special program is running and to be able to redirect the standard output elsewhere.
Thanks! Stupid question - how do you toolchain pylint? I am using the code.pybricks.com now, that is awesome, yet pylint and auto format is not av.
I have my sources on Linux Mint and copy them into pybricks. So I have pylint on linux.
Thanks for working on this.
Any hints how I can channel the terminal output to the dialog?
We will either need a new event type or status bit to indicate that this special program is running and to be able to redirect the standard output elsewhere.
I think this could be covered by the suggested approach above. Only this program would be started by the dialog. And it need not use standard output.
Of course you're welcome to work on it as you like, but it may be a bit soon to commit to a particular approach :)
I would like to make some time to clean up https://github.com/pybricks/pybricks-micropython/pull/246 and keep #1708 in mind along the way. See also the recent note there about sending data back to the IDE.
Here is my first naive and wacky implemetation. https://github.com/afarago/pybricks-code/tree/experiment-hubcentral
And the demo video: https://photos.google.com/share/AF1QipO47B0qloVnoIxXApMJsUyUWYs81w2MKyqVvzJEJNopYzwn7MLgs8BmvyrR1Po62A/photo/AF1QipOcD6qFdfQ7nt0w9yYZjpAzdFLW93JLpzBsn8MT?key=ZTZXaFk4c2pYOGhHY3gzZHEyQlB0RlNLOE1mLVFB
Will keep on experimenting with the HostBuffer approach, currently just hacked and replicated the terminal stream.
I think I am pretty much done with the base functional implementation.
Let me know how we could move on with the next steps. Also any review comments are welcome, to get code compatible/compliant.
- https://github.com/afarago/pybricks-code/tree/experiment-hubcentral
- https://gist.github.com/afarago/350a931992e05c423860f4b25ea5c6a2
What is definitely left
- HostBuffer / AppData usage
- firmware inclusion of the hubcenter code (or similar)
- IDE to activate the hubcenter mode on the connected brick
- current implementation follows dark mode, need to make it cross compatible for both modes
Whoa, that looks amazing! I think it's going to take quite a bit of time before I can get to this, but I'd really like to merge something like this eventually!
Here's a tangentially related idea that could perhaps be its own topic in the future. I'm just mentioning it now in the context of potentially keeping the communication schemes very generic.
In addition to a port view, we could have a live variable view for the currently running script. Since we already require variables to be declared in the setup section, we could quite easily add > option on their setup block to activate monitoring it. Doing so could generate an async background task that sends that value as app data when it changes.
In the same vein, you can think of a way to highlight the current block or stack.
FWIW, I am currently working on the AppData feature in https://github.com/pybricks/pybricks-micropython/pull/246, as well as a generalized way to start programs.
Currently we have:
pbio_error_t pbsys_user_program_start_program(void);
pbio_error_t pbsys_user_program_start_repl(void);
These commands will be internally generalized to:
pbio_error_t pbsys_main_program_request_start(pbsys_main_program_type_t type, uint32_t id);
At the protocol level, I would like to propose to rename PBIO_PYBRICKS_COMMAND_START_REPL to PBIO_PYBRICKS_COMMAND_START_BUILTIN_PROGRAM and give it an identifier payload. The payload can be optional for backwards compatibility.
Likewise, PBIO_PYBRICKS_COMMAND_START_USER_PROGRAM can get an optional identifier payload too so we don't have to modify again at some point in the future.
The other option would be to make these new, separate commands.
Additionally an idea: If we are able to reliably and easily transmit what is connected to the hub, that could enable the IDE to generate the template with all the peripherials already initalized, e.g motors, color sensor on the righr port both in text and blocks mode.
That would speed up the onboarding process for newbies.
On an implementation level of the Python side, here is an experiment with device-specific output:
The port and type ID are standard and separated by a tab. The remainder is device specific plain-text we can show in Pybricks Code with limited further processing, if any. The device type ID is just so we can show the right picture.
That way, we don't need to re-implement things such as values[0] = forcevalue if forcevalue > 0 else self.device.read(1)[0] # FORCE or TOUCH since the firmware is already doing those things for us. This also ensures we show exactly the same values as users would see in their own code.
Unknown devices can show their default mode data.
The remainder of the code is a small runloop that pushes these generators along when something gets plugged in. I'll add it to the proposal in #254.
In the sensor UI previewed by Attila above, relevant sensors can have a dropdown (or make the sensor icon clickable) to select a mode. These need not be all 10+ modes, but just the relevant ones in our API:
For example, the boost color and distance sensor might have:
- Reflected light intensity and color
- Ambient light intensity
- Distance
The list of items to populate the dropdown with comes from the hub script as plain text, too. The user selection (index) can be sent to the hub with app data, so it knows which sensor values to send:
For example:
# BOOST Color and Distance Sensor
def update_color_and_distance_sensor(port, type_id, mode):
sensor = ColorDistanceSensor(port)
# This data populates a mode dropdown in the IDE.
modes = [
"Reflected light intensity and color",
"Ambient light intensity",
"Distance"
]
yield "\t".join([str(port)] + modes)
# Report the sensor data as preformatted text.
while True:
if mode == 0:
hsv = sensor.hsv()
ref = sensor.reflection()
data = f"h={hsv.h}°, s={hsv.s}%, v={hsv.v}%, r={ref}%"
elif mode == 1:
data = f"a={sensor.ambient()}%"
else:
data = f"d={sensor.distance()}%"
yield f"{port}\t{type_id}\t{data}"
So for the IDE, "incoming data" looks as follows:
Port.A --
Port.B 75 -115°
Port.C 61 h=120°, s=7%, v=17%, r=9%
Port.D --
Port.E --
Port.F --
Port.A --
Port.B 75 -115°
Port.C 61 h=60°, s=7%, v=17%, r=9%
Port.D --
Port.E --
Port.F --
Port.A 37 modes Reflected light intensity and color Ambient light intensity Distance
Port.A 37 h=0°, s=0%, v=0%, r=0%
Port.B 75 -115°
Port.C 61 h=0°, s=0%, v=19%, r=10%
Port.D --
Port.E --
Port.F --
Multiple whole lines (separated at "\n") may be sent in one package, but lines are never broken up. Within each line, the itself of interest are "\t" separated.
In this example, B and C were already plugged in, and then a sensor was plugged into port A.
Even though this will only run while no other program is running, it would still be useful to design the UI such that it can be viewed alongside the editor.
Typically, you'd want to use port view to find things like color values or threshold, and you want to enter those into your code as you go. It wouldn't be ideal if you had to go back and forth several times for each value.
Absolutely right point!
I am happy to contribute to the core content, yet the reason for chosing the dialog was that it had less impact on the overall structure of the app, but definitely could make a lot more sense as a side panel. Either with spltters or even float in.
If @dlech would modify the app frame (splitters/etc) I can come up with a full implementation for the port view if that helps.
(Later on: It has also the potential to have reactive contents, so if it is resizeable It might even have different levels of granularity.)
After exchanging ideas last week, the following seems to be the direction to go in:
Port view can be started by clicking on the hub status
Port view has its own dialog
- We'll use dedicated dialog like this, instead of sideview as suggested above. A dedicated view makes more sense within the overal UX and technical constraints.
- Opening the dialog starts the Port Control program on the hub.
- Closing it stops this program.
Port view style features actual components
Since Pybricks supports much more than just one LEGO set, it could make sense to display what is connected instead of abstract icons. We can render a few images of pre-routed cables without devices, and separate images for the devices and hubs so we can dynamically show the correct view. A visually rotating motor with a red pointer as shown below would be worthwhile adding as well.
All images can be rendered to respect the LEGO unit grid for easy placement.
Implementation
Most of the required tools like the AppData class are now merged into the master firmware branch. Further changes to the port view program are welcomed.
As a first step, it would be awesome to bring @afarago 's suggestion above back to life with the above suggestions.
Until we have rendered everything, we can use these images for development.
Merging in some ideas posted elsewhere...
I'd love to see two buttons added to the web IDE. (...) 'turn the hub off'. While turning the hub on is very easy (just click the button) it takes a loooong press to turn it off and you have to get physically to the hub. So more often than not I fire REPL up and execute TechnicHub().system.shutdown() instead. I'd love to just have an IDE button that'd do just that. Shut the hub remotely down.
Originally posted by @biasedlogic in https://github.com/pybricks/support/issues/1830#issuecomment-2357716925
If there was ever a place for it, maybe the hub/port view can have a shutdown button somewhere in its dialog.
Within the python program can I somehow detect whether the hub is connected to pybricks-code IDE or not?
I would like to stop the portview app if the IDE is closed. (Upon disconnect I am able to stop the program from the pybricks-code IDE now.)
I tried stdio / keyboard poll / print as a workaround, yet they handle a failure (no terminal/remote app) pretty well.
Within the python program can I somehow detect whether the hub is connected to pybricks-code IDE or not?
Good timing! This is being worked on / brainstormed in https://github.com/pybricks/support/issues/1496#issuecomment-2395009700