Re: [Boost-testing] regression.py/Boost.Build v2 issues

From: Vladimir Prus (ghost_at_[hidden])
Date: 2006-02-26 03:13:02

Reece Dunn wrote:

> I have found another issue with pregression.py. After doing a V2 run,
> status contains directories and filenames that have a character that is
> displayed as a square (unrepresentable). I.e., it is testing for Unicode
> characters. However, when regression.py is removing a previous boost run,
> I get the following error:
> 
> # Getting sources (2006-02-23T07:58:08Z)...
> # Downloading
> "http://www.meta-comm.com/engineering/boost/snapshot/boost-CVS-HEA
> D.tar.bz2" to "C:\devel\regression"...
> # Looking for old unpacked archives...
> # Unpacking boost tarball
> # ("C:\devel\regression\boost-CVS-HEAD.tar.bz2")...
> #     Unpacked into directory
> "C:\devel\regression\boost-CVS-HEAD-06-02-23-0807"
> 
> # Deleting "C:\devel\regression\boost" directory...
> # get_tarball failed with message "[Errno 22] Invalid argument:
> 'C:\\devel\\regr
> ession\\boost\\status\\?'"
> 
> Here, the Unicode character has been replaced by a '?' character and is
> failing when calling rmtree. I have tracked it down to the
>    shutil.rmtree( path )
> call trying to remove the directory containing a Unicode character.

Strange. Then stock version of regression.py should have the same problem as
I did not change anything. Maybe it's just was never detected.

> Another question: I am using cygwin bash to run the regression tools, but
> running a Windows version of python, so is it possible to detect this
> situation and use the unix/cygwin tools?

Can you clarify? 

- Volodya