Closed WnP closed 9 years ago
done
@WnP do you have some test cases that caused failure in the previous implementation?
I'd love to have a better way to test this.
@lukasschwab
% stackit -e "python -c 'print(foo)'"
Searching for: python: can't open file '/home/steeve/stackit/-c': [Errno 2] No such file or directory ...
. . .
should search for NameError: name 'foo' is not defined
not python: can't open file '/home/steeve/stackit/-c': [Errno 2] No such file or directory
for example, and all other command line with arguments that returns errors
Great, I'll test with those examples and my old test set and hopefully merge by EOD.
@lukasschwab corrected ;-)
@WnP Looks great! I tested with your example, and also with my old set of tests for stackit -e
(Python scripts that generate errors), and tested the new "no error" message with bash commands (e.g. ls
) and Python scripts that don't generate errors.
Merging. Thanks!
@schwartzmx Check this out––let me know if it solves the issue you brought up on PR #6. I think it should...
@lukasschwab @WnP I know Windows isn't currently "supported". But this is the type of anomalies I am talking about, they may not be solvable the way different shells and programs output stack traces and errors. For example in Powershell (With output questions abbreviated ...
:
C:\stackit\stackit [master]> python .\stackit_core.py -e "ls -adfa"
Searching for: operable program or batch file....
Tags:
...
Enter m for more, a question number to select, or q to quit: q
C:\stackit\stackit [master]> ls -afda
Get-ChildItem : A parameter cannot be found that matches parameter name 'afda'.
At line:1 char:4
+ ls -afda
+ ~~~~~
+ CategoryInfo : InvalidArgument: (:) [Get-ChildItem], ParameterBindingException
+ FullyQualifiedErrorId : NamedParameterNotFound,Microsoft.PowerShell.Commands.GetChildItemCommand
The search that was done has nothing to do with the actual error. I don't know if this is solvable for everything. But I think this is regardless of just Windows. Just figured I'd bring this up, to clarify what I meant in the earlier thread.
It does work for things (that I've tested) on Unix.
Phil
@schwartzmx I don't have Windows to be make some test, I gess the problem come from the fact that errors are not redirected to stderr
could you show me this test script output to better understand the problem?
Maybe we should redirect
stdout
to/dev/null
, we could do that by settingPopen
'sstdout
option to os.devnull