Open GoogleCodeExporter opened 8 years ago
Thanks for the report. I've opened an internal issue to track this.
Original comment by dlor...@google.com
on 1 Feb 2016 at 4:39
Also encountered this on Windows 10, likely due to an accented character in the
CWD path.
Original comment by jes...@google.com
on 2 Feb 2016 at 1:26
Can you run again with --format=flattened and post the last few lines of
standard output before the ERROR: message. Flattened flushes line by line to
the terminal and it will give a clue to the key that has the encoding problem
(the key that would print after the last one that displayed).
Original comment by gsfow...@google.com
on 2 Feb 2016 at 4:18
Sure:
TKE:~ mbishop$ gcloud compute instances describe console2 --format=flattened
canIpForward: True
cpuPlatform: Intel Haswell
creationTimestamp: 2016-01-29T02:02:18.329-08:00
description:
disks[0].autoDelete: True
disks[0].boot: True
disks[0].deviceName: console2
disks[0].index: 0
disks[0].interface: SCSI
disks[0].kind: compute#attachedDisk
disks[0].mode: READ_WRITE
disks[0].source:
https://www.googleapis.com/compute/v1/projects/xxxxxxx/zones/us-central1-b/disks
/console2
disks[0].type: PERSISTENT
id: xxxxxxxx
kind: compute#instance
machineType:
https://www.googleapis.com/compute/v1/projects/xxxxxxx1/zones/us-central1-b/mach
ineTypes/custom (6 vCPU, 16.00 GiB)
metadata.fingerprint: xxxxx
metadata.items[0].key: startup-script
ERROR: gcloud crashed (UnicodeEncodeError): 'ascii' codec can't encode
character u'\u201c' in position 398: ordinal not in range(128)
Original comment by mbis...@logitbot.com
on 2 Feb 2016 at 6:16
Its a unicode char in the startup script: '“' == 'LEFT DOUBLE QUOTATION
MARK'. It should probably be the plain old ASCII double quote '"'.
This is definitely a describe command unicode bug, but there may be another
problem.
Did you cut and paste the startup script from a web page? If its intended to be
a shell or cmd.exe double quote character then the script may be running
something other than intended.
Original comment by gsfow...@google.com
on 2 Feb 2016 at 8:04
Ahh, the startup script. Yes, come to thinks of it I believe I did c&p from
either an RTF doc or a web page, def somewhere that could have been non-ascii.
I'll make sure it's all ASCII and give it another go.
Thanks for the help, good catch!
Original comment by mbis...@logitbot.com
on 2 Feb 2016 at 8:50
Thanks for following up. Interesting that after we fix our encoding bug
cut&oaste problems like this may be harder to detect.
Original comment by gsfow...@google.com
on 2 Feb 2016 at 10:01
May want to keep an undocumented flag to "un-fix" the bug to help find these
guys in the future. :)
Original comment by mbis...@logitbot.com
on 2 Feb 2016 at 10:15
Yeah, unintended consequences of being correct. Once we release fixes for a few
unicode decoding bugs I'll post a filter that can weed out bugs like this, no
special flag required.
Original comment by gsfow...@google.com
on 5 Feb 2016 at 2:38
Original issue reported on code.google.com by
mbis...@logitbot.com
on 29 Jan 2016 at 10:27