[Feature] Remotely starting a stored program with pybricksdev

[Novakasa]

Is your feature request related to a problem? Please describe.
I am incorporating multiple PoweredUp Hubs in a larger project which is controlled centrally on a PC using pybricksdev. When running it, I have to run BLEPUPConnection.run() for each Hub, which for me takes about 10 seconds (see #283). Given that we have the possibility to store a program on the hub permanently, it would be a nice feature to be able to start that program remotely using pybricksdev, so the program will not have to be copied to the hub every time I am testing my larger project. Even when #283 is addressed, each download and run operation takes about 2-3 seconds for me even with Pybricks Code, so I still think that this would be a useful addition to pybricksdev especially if you are using scripts of large sizes, which might take a bit longer to copy still.

Describe the solution you'd like
I can imagine adjusting the API by seperating copying and starting a program like so:

hub = BLEPUPConnection()
await hub.connect(my_address)
await hub.copy_script("my_script.py")
await hub.run_script("my_script.py")

And this should also work when ommitting copy_script when a program was already stored permanently when flashing the firmware.
If there might ever be the possiblity of storing multiple scripts on the hub, one could introduce a separate name parameter for copy_script and run_script so the scripts can be identified.

[laurensvalk]

It would be a nice feature to be able to start that program remotely using pybricksdev, so the program will not have to be copied to the hub every time I am testing my larger project.

Although seemingly unrelated, this would be solved by #408, by running a script that just does import main.

And this is perhaps even more useful than a special start command, since you'd be completely flexible in passing additional commands. You could upload the same main.py to all trains, and then start each with its own parameters as needed:

from main import SmartTrain

train = SmartTrain(param1, param2)
train.start()

[laurensvalk]

This is now theoretically possible with pybricks/pybricks-micropython#109 and subsequent PRs. Sending program size 0 is the same as pressing the button.

Although it might be nicer to make a command for it just like we have the stop command. Then we can get rid of the special size to trigger the REPL as well.

[Novakasa]

This is now theoretically possible with pybricks/pybricks-micropython#109 and subsequent PRs. Sending program size 0 is the same as pressing the button.

Could that currently work using pybricksdev?

[dlech]

Although it might be nicer to make a command for it just like we have the stop command. Then we can get rid of the special size to trigger the REPL as well.

This has been done in the firmware, so it is possible now. There just isn't a high-level API for it in pybricksdev. It should be easy enough to split the download and run function into two functions.

[Novakasa]

Oh yeah, I see it now, I have no idea why I missed this earlier

[Novakasa]

If I understand correctly, the new function start_user_program indeed does what is requested here, however it seems that it doesn't really replace the old run function, since a lot of setup and state management for stdin, stdout, logfiles, etc still is only done in run (which always redownloads the program. Is start_user_program supposed to be called in isolation?

For me, the ideal API would be to have run but just have the download be optional.