On 10/26/2010 05:27 PM, Stefan Berger wrote:
On 10/26/2010 07:08 PM, Eric Blake wrote:
Sorry for sounding so depressing; I'm very pleased to see your efforts in providing tests for the code you've written. Even though this test is intended to be skipped on non-Linux, you still have to worry about merely parsing through the test on other platforms like Solaris, where /bin/sh won't understand the bash-isms and where /bin/bash is not guaranteed to exist. But if we decide that requiring the presence of /bin/bash is acceptable for the TCK, then a lot of my review becomes irrelevant; but my comments about your mkstemp replacement being insecure are still applicable even in that case.
Ok. Well, I hope for bash then...
IRC verdict - Dan and I are both okay with assuming /bin/bash exists. So, that just leaves cleaning up the temporary file management. Would you like me to tackle that as an incremental diff on top of your original patch? -- Eric Blake eblake@redhat.com +1-801-349-2682 Libvirt virtualization library http://libvirt.org