[master] AttributeError in filter_traceback: 'str' object has no attribute 'relto'
#995
Labels
topic: reporting
related to terminal output and user-facing messages and errors
type: bug
problem that needs to be addressed
Milestone
Comments
|
With pytest 2.7.2 (instead of master), I see this error during setup of tests, |
blueyed
changed the title
filter_traceback: 'str' object has no attribute 'relto'filter_traceback: 'str' object has no attribute 'relto'
|
Hi @blueyed, Could you try the patch at https://bitbucket.org/pytest-dev/py/pull-requests/31? Seems related. |
|
I'm having the exact same error and the patch did not work. Python 3.5.0 running in virtualenv, on Mac OS X 10.11 Beta 6 |
|
@blueyed @ColtonProvias Do you have a minimal example or a public project where this happens? |
|
Perhaps it would be best for pytest to be on the safe side in this situation and guarantee that entry.path is always a entry_path = py.path.local(entry.path)
return entry_path != cutdir1 and not entry_path.relto(cutdir2)Either way, it would be nice to get a minimal example, or at least a public project where the error is easily reproducible. 😅 |
|
I'm starting to see this now as well, I'll try to get some more debug info: entry = " File '<string>':2 in installed_client\n ???\n"Where installed_client is a fixture in which the exception was raised. And here is some of the stack: |
|
I believe the issue is using the decorator package (https://pypi.python.org/pypi/decorator). I am using it to preserve certain fixtures signatures while attaching an additional decorator to it. From the documentation of decorator: Calling f() will give you a ZeroDivisionError, but since the function is decorated the traceback will be longer:
>>> f()
Traceback (most recent call last):
...
File "<string>", line 2, in f
File "<doctest __main__[22]>", line 4, in trace
return f(*args, **kw)
File "<doctest __main__[51]>", line 3, in f
1/0
ZeroDivisionError: ...
|
|
In my case, running on python 3.4, the After upgrading my mock to ver 1.3 I was also able to use |
|
Recently we had a session level fixture that started failing. The error message in 2.7.3 was correct and helpful (with the complete description "ERROR as setup of ...."), but in 2.8.0 we just had a stack trace ending with Simply correcting the external condition that was causing the fixture to fail (which is actually to create some particular directory) fixes the problem. That fixture uses a decorator: @pytest.fixture(scope='session')
def sc():
spark_home = os.environ['SPARK_HOME']
sys.path.insert(0, os.path.join(spark_home, 'python'))
sys.path.insert(0, os.path.join(spark_home, 'python/lib/py4j-0.8.2.1-src.zip'))
import pyspark
return pyspark.SparkContext('local', 'tests') |
blueyed
added
type: bug
|
I got the same as @deuxpi, an error from code being called inside a fixture is hidden by this "AttributeError: 'str' object has no attribute 'relto'" error. Sorry nothing reproducible, but FWIW my structure is: The triggering error in my app (discovered when examining outside of pytest) is: |
|
Here is a simple gist that will trigger the error: https://gist.github.com/astraw38/e485eb89cf6cf61c7be9 The 'traceback.cut()' method does not filter out the _pytest internal code if you have an anonymous function in a different file. |
nicoddemus
added a commit
to nicoddemus/pytest
that referenced
this issue
Sep 30, 2015
…rated by an exec() statement Fix pytest-dev#995
|
Thanks @astraw38, that helped me narrow down a simple test case and fix the issue. 😄 |
RonnyPfannschmidt
added a commit
that referenced
this issue
Oct 1, 2015
Fix internal error when filtering tracebacks where one entry was generated by an exec() statement Fixes #995
|
We're still seeing this error message running against pytest 2.8.2, so it may not be fixed. |
|
Here is the traceback we see: Interestingly, I'm unable to replicate the issue on my local workstation which is OS X and Python 2.7.10. I'll continue to troubleshoot until I can narrow down a cause. In the meantime, can you suggest a way to inspect the failure where it does exist? |
Can you use |
|
@nicoddemus Thanks for the hint. Yes. It works in the latest master. Thanks. |
|
Will there be a release with this fix anytime soon? I keep running into this. |
|
This should have been out in 2.8.3... which version are you using? |
|
Sorry, was running an old version by accident! |
|
Np! 😄 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
I am seeing the following error suddenly:
Normally
entry.pathis of type<class 'py._path.local.LocalPath'>, but in this case it is not.Full traceback:
The text was updated successfully, but these errors were encountered: