Reviewer needed - fix for regression test LastErrorString

Alan Bateman Alan.Bateman at
Sun May 15 18:55:20 UTC 2011

Pavel Tisnovsky wrote:
> :
> I changed this regression test so it does not need to use shell script
> at all because now it creates all required files itself.
> Here is new webrev:
> I've also ran this test on Windows XP SP2 and it seems to works fine.
> Output of this test is stored in an attachment [stderr_windows.log].
> Please note that I found only Win XP configured with Czech locale so
> some system messages might me incomprehensible, but these messages are
> correct ("Přístup byl odepřen" == "Access denied" etc.)
Thanks for changing the approach to avoid the shell script. Also thanks 
for testing it on a Windows machine, unpalatable as it might be to have 
to do that.

If I read the test correctly then it will now leave unreadable and 
unwritable turds in the working directory. I see you clean up on re-run 
but it would be nicer if the test just cleaned up after itself.

It would be nice to remove the attempt access of c:/pagefile.sys and 
z:/fooBAR/baz/GORP while you are there. That will simplify preparePaths 
as the runUnderWindows parameter won't be needed.  As the work directory 
is the current directory then workDirParam isn't really needed either if 
you'd like to get rid of that too.

Minor comment is that the comment on line 144 references the shell 
script that isn't in this version.  Also, is the 
@SuppressWarnings("unused") at line 55 needed?


More information about the core-libs-dev mailing list