"undefined" in stack trace
Artem
Posted on
Dec 15 2014
When error.stackTrace.methodName is set to null while other fields are present, Raygun web app shows "undefined" for all fields:
File "undefined", line undefined, in undefined
undefined
Here is a sample JSON sent to Raygun:
{
"userCustomData": {
"LOCALE": "en_US, cp1252"
},
"client": {
"version": "2.1.0",
"name": "raygun4py",
"clientUrl": "https://github.com/MindscapeHQ/raygun4py"
},
"error": {
"className": "__init__",
"stackTrace": [
{
"className": "<module>",
"methodName": null,
"fileName": "C:\\Program Files (x86)\\JetBrains\\PyCharm 3.4.1\\helpers\\pydev\\pydevd.py",
"lineNumber": 1733
},
{
"className": "run",
"methodName": null,
"fileName": "C:\\Program Files (x86)\\JetBrains\\PyCharm 3.4.1\\helpers\\pydev\\pydevd.py",
"lineNumber": 1226
},
{
"className": "execfile",
"methodName": null,
"fileName": "C:\\Program Files (x86)\\JetBrains\\PyCharm 3.4.1\\helpers\\pydev\\_pydev_execfile.py",
"lineNumber": 38
},
{
"className": "<module>",
"methodName": null,
"fileName": "B:/Development/projects/main.py",
"lineNumber": 11
}
],
"message": "Sample Exception",
"data": ""
},
"environment": {
"oSVersion": "Windows 8",
"processorCount": 8,
"cpu": "Intel64 Family 6 Model 58 Stepping 9, GenuineIntel",
"architecture": "64bit"
},
"machineName": "a-martynovich",
"version": "42-pre",
"tags": [
"win32"
]
}
And here's what I see on Raygun page:
Sample Exception
File "undefined", line undefined, in undefined
undefined
File "undefined", line undefined, in undefined
undefined
File "undefined", line undefined, in undefined
undefined
File "undefined", line undefined, in undefined
undefined
Setting methodName to "" (empty string) has the same effect.
I have a "null" methodName because that's what Python traceback contains when I run optimized (compiled) .pyo files. I think it is a Raygun bug because the Raygun documentation doesn't say that all traceback fields will suddenly turn "undefined" when one field is null.
John-Daniel Trask
Raygun
Posted on
Dec 15 2014
Hi Artem,
Thank you for posting this. One of our team will take a look at this and post back here with what they find. Hopefully it's a simple bug that we can resolve quickly for you.
Kind regards,
John-Daniel Trask
Co-founder
Mindscape
Kentzo
Posted on
Dec 15 2014
This is also very important to us!
Looks like it's the issue in the rendering code since old reports are broken as well.
Callum
Posted on
Dec 16 2014
Thanks for alerting us of this, guys - there was indeed a regression in a part of the rendering code. A fix has now been applied and your error views should be back to normal. Apologies for any inconvenience.
Regards,
Callum Gavin
Mindscape Limited