Roku Developer Program

[destruk]

Channel name -- dev channels
Channel version -- internal/loaded dev channels

When adding a stop command to break program flow and enter the debugger - and also when pressing Control-C in the Telnet window, half the time the roku exits the running channel and the debugger window fails to respond. In other words, half the time when the program flow is stopped in a routine, or during operation, the roku exits the channel code entirely and returns to the home screen menu

What are the steps to reproduce? -- Run a dev channel, have it execute a stop command in the code or press Control-C with the debugger telnet window open
Is it reproducible (how many times out of 5 does this happen)? -- 3 out of 5
What is the serial number of the box? -- 13A16X003375
What software version is the box running? -- 5.0 Build 8043

Is the box connected to the internet via wireless or a wired connection? -- wifi
What signal strength is the box reporting if connected to the internet via Wifi? -- excellent
What type of internet service does the user have? -- 40MBit DSL
How is the box connected to the TV -- HDMI
What resolution is being used? -- 1920x1080

[TheEndless]

For me, this only seems to happen the first time after the initial side-load. So, if you exit the channel and re-launch it as part of your side-loading procedure, you should be able to break into it from that point on without it exiting.

[destruk]

Yeah, it is frustrating when trying to do a quick debug of some code. Sometimes it works.

[EnTerr]

THIS!
I just lost an hour of work because of this dev.environment bug and am fuming.
Can someone Roku* have a look please - is this logged as a bug? What are expectations of it getting fixed?

And here is how i got screwed by it today: i have been running my simulation on Roku and after about an hour noticed there are some boundary issues/artefacts. Aha - now if i look at the raw data i will know what's wrong - just have to interrupt it and print an array - so i press ctrl-C at iteration #52063 and... i am fracked, backtrace prints but program exits and there is nothing i can do but start from the beginning. :cry:

I think everyone that develops for B/S know the behavior and we kind of go along with it since "you just have to restart your program once more". And once more. And once more. What is interesting is that it happens EVERY OTHER TIME, as a metronome - tick-tock, tick-tock. Meaning if you interrupt program at time N and console works, next start N+1 a break will exit to main menu, then at N+1 will break into console and so on. It feels like something reliable that easily can be tracked down and fixed, some boolean kind of condition.

Any other developers tormented by this, can you add your voice (and horror stories) here, hopefully to get some attention to the issue?

[belltown]

This has been tormenting me too. Every other time exactly the channel exits the debugger. I'm now in the habit of running the channel twice whenever I want to debug something, the first time knowing it will exit without letting me debug, the second time is when I get to do my actual debugging. It doesn't seem to be related to whether I've just side-loaded the channel or whether I've run it from the home screen.

[renojim]

Yeah, it's ridiculous that this behavior still exists after all these years, but I don't see the same behavior as you do. For me, it seems to almost always fail the first time I hit Ctrl-C (or my code hits a stop), but after that it seems to work ok. I've gotten into the habit of starting my development channel and hitting Ctrl-C immediately and then restarting the channel and repeating until Ctrl-C works properly. After that, I can't say that I have any problems, but I haven't been doing a whole lot of development lately.

-JT