Unicode safety

[Adam Nelson]

It appears that we're using the ascii codec for interrogating the frame:

{{{ #!python

ERROR:root:Request watchdog failed Traceback (most recent call last): File "/var/www/example-env/src/dogslow/dogslow/init.py", line 128, in peek output += stack(frame, with_locals=True) UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 46514: ordinal not in range(128) }}}

[Erik van Zijst]

Yeah, that doesn't look right.

[RussF]

Just hit the same problem.

[TzongHao Chen]

Inserting this line at the top of init.py would fix the problem:

# -*- coding: utf-8 -*-

[Erik van Zijst]

Which init.py? Dogslow's? I don't think that'll help as Dogslow is not parsing its own source file.

I suspect there's a string on the stack that is non-ascii and I'm implicitly decoding it as such.

[Eleni Lixourioti]

Is there any progress on this one? I seem to be having the same issue...

[Erik van Zijst]

25f884776471 should address this for now.

[Erik van Zijst]

Note that this is not a proper solution. For a thorough breakdown of the problem, see https://bitbucket.org/evzijst/dogslow/pull-request/10/fixed-string-concatenation/diff#comment-202906