Application exception output to serial (over USB) appears truncated due to buffering issue

[kevinjwalters]

My near-perfect code has just spat an exception. I'm running 4.0.0-beta.3-4-g186e31591-dirty which 4.0.0 beta 3 with a minor tweak. The exception has half-appeared on my console (PuTTY on Windows) from the CPX over USB:

Press any key to enter the REPL. Use CTRL-D to reload.soft reboot

Auto-reload is on. Simply save files over USB to run them or enter REPL to disable.
code.py output:
Shared couter PWM failure I - trying in reverse order
Shared couter PWM failure II - soft/hard reset suggested
Ready to play
Traceback (most recent call last):
  File "code.py", line

There's a single space after line - the absence of the rest of the text on the line makes it very apparent it's truncated. When I press the enter key and the rest appears leaving it looking like this:

Press any key to enter the REPL. Use CTRL-D to reload.soft reboot

Auto-reload is on. Simply save files over USB to run them or enter REPL to disable.
code.py output:
Shared couter PWM failure I - trying in reverse order
Shared couter PWM failure II - soft/hard reset suggested
Ready to play
Traceback (most recent call last):
  File "code.py", line 199, in <module>
TypeError: list indices must be integers, not tuple



Press any key to enter the REPL. Use CTRL-D to reload.
Adafruit CircuitPython 4.0.0-beta.3-4-g186e31591-dirty on 2019-03-11; Adafruit CircuitPlayground Express with samd21g18

I've never noticed that before and I use this setup a lot. It appears fully reproducible for this particular code/error combo. Are there differences in the way 4.0.0 vs previous versions writes to serial or how it reports exceptions?

[tannewt]

4.x uses a different USB stack than 3.x so it's likely a bug in TinyUSB. Maybe @hathach will take a look.

[kevinjwalters]

For what it's worth, just noticed this occurs with 4.0.0 beta 5 too. Only about 1 in 20 of my errors does this, there's something data (or timing?) dependent about this.

[notro]

It's probably the same problem I have run into #1342 (comment)

[kevinjwalters]

FYI, In my case the data will pop out if you tap return (i.e. send some data in oppposite direction). The terminal session is not disconnected.

[tannewt]

I wasn't able to reproduce this. @kevinjwalters can you please post everything needed to reproduce this? Thanks!

I tried this code because it prints long strings:

import time

for i in range(100):
    print("h"*i)

for x in range(4):
    print("wowo")


print("hellow")

print("woowza")

print("blef")

time.sleep(0.1)


for i in range(100, 500):
    print(("h",)*i)

[kevinjwalters]

I've only noticed it on certain exception output, not to say it doesn't occur on others, of course.

My first example got zapped when I encountered the #1643 bug :( I may have preserved another, let me go digging.

[kevinjwalters]

This did it on the 3rd attempt. Each time I was using up arrow/return to recall and execute the previous command in PuTTY. It's not fully reproducible though:

>>> raise AttributeError("abcdefghij" * 12)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
AttributeError: abcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghij
>>> raise AttributeError("abcdefghij" * 12)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
AttributeError: abcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghij
>>> raise AttributeError("abcdefghij" * 12)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
AttributeError: abcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghij

[kevinjwalters]

Here's one showing where it stops (until any key (tried return, tried c) is pressed which then nudges the rest of it out):

>>> raise AttributeError("abcdefghij" * 12)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
AttributeError

It's also possible this is triggered by some very subtle behaviour across USB to terminal...

[kevinjwalters]

I'd suggest leaving this until I can get a more reproducible case.