scripts like webkit-patch will fail on Windows due to an error in executive.py. Based on the description in https://docs.python.org/2/library/subprocess.html "Note that on Windows, you cannot set close_fds to true and also redirect the standard handles by setting stdin, stdout or stderr." Setting close_fds to False on Windows solved the issue.
Created attachment 307059 [details] Patch
Can someone take a look at this?
I think setting close_fds to true will work if you run Python 2.7.10 or later.
(In reply to Per Arne Vollan from comment #3) > I think setting close_fds to true will work if you run Python 2.7.10 or > later. Thanks for the reply. I'm running python 2.7.13, which I believe should work, and in fact it doesn't. >python --version Python 2.7.13 Besides, the following line is from python 2.7.13 doc: > Note that on Windows, you cannot set close_fds to true and also redirect the standard handles by setting stdin, stdout or stderr. https://docs.python.org/2/library/subprocess.html
(In reply to Bill Ming from comment #4) > (In reply to Per Arne Vollan from comment #3) > > I think setting close_fds to true will work if you run Python 2.7.10 or > > later. > > Thanks for the reply. > > I'm running python 2.7.13, which I believe should work, and in fact it > doesn't. > > >python --version > Python 2.7.13 > > Besides, the following line is from python 2.7.13 doc: > > > Note that on Windows, you cannot set close_fds to true and also redirect the standard handles by setting stdin, stdout or stderr. > > https://docs.python.org/2/library/subprocess.html That is strange, I don't seem to have this issue with Cygwin Python 2.7.10. Are you running Cygwin?
No, I'm using the official python installation package for windows. Maybe in cygwin the stdio is treated differently than CMD.
(In reply to Bill Ming from comment #6) > No, I'm using the official python installation package for windows. > > Maybe in cygwin the stdio is treated differently than CMD. It makes sense that stdio behavior between Cygwin and native Windows Python could be very different. I think this change seems fine, as long as it's limited to non-cygwin builds.
Comment on attachment 307059 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=307059&action=review > Tools/Scripts/webkitpy/common/system/executive.py:100 > + return True This seems reasonable for Windows-specific behavior.
Comment on attachment 307059 [details] Patch Clearing flags on attachment: 307059 Committed r215523: <http://trac.webkit.org/changeset/215523>
All reviewed patches have been landed. Closing bug.