-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/22273/#review45078
-----------------------------------------------------------


This change is not sufficient.

It only works on the systems where it works because the shell association 
between *.py and the python interpreter is already setup.

It is very possible that it won't be - for instance this will happen on a 
system that has *multiple* installed programs that claim that they can handle 
*.py.

In this case even telling python to use the shell to run *.py will fail (and 
will not give a useful meaningful error report - I know I've been there!).

The only way to make this work reliably is to make the test script itself run 
*.py using python. Fortunately, I think there are only 2 cases of external 
programs to run native executables and python scripts, so we could look for the 
.py in the script and special case it. 

- Andrew Stitcher


On June 5, 2014, 6:14 p.m., Chug Rolke wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/22273/
> -----------------------------------------------------------
> 
> (Updated June 5, 2014, 6:14 p.m.)
> 
> 
> Review request for qpid and Rafael Schloming.
> 
> 
> Bugs: PROTON-595
>     https://issues.apache.org/jira/browse/PROTON-595
> 
> 
> Repository: qpid
> 
> 
> Description
> -------
> 
> Popen(file.py, ...) needs to run with shell=True on windows. On Linux the 
> tests hang when shell=True. This patch compromises by turning on the shell 
> arg when running on windows.
> 
> Change the arg list to run Popen(python, file.py, ...) touches wider ranges 
> of source code with some risk and with no obvious benefit.
> 
> 
> Diffs
> -----
> 
>   proton/trunk/tests/python/proton_tests/common.py 1600676 
> 
> Diff: https://reviews.apache.org/r/22273/diff/
> 
> 
> Testing
> -------
> 
> Passes tests on linux and windows.
> 
> 
> Thanks,
> 
> Chug Rolke
> 
>

Reply via email to