Pip provider fails on RedHat osfamily when pip not installed via RPM (Could not locate the pip command)

[Michael O'Dea]

In resolving bug 15980, the pip package provider was changed to recognize the RedHat osfamily fact and use the executable name "pip-python" in those environments.  This does not take into account environments where pip has been installed via easy_install or by custom rpm providers.  In those environments (such as mine) this osfamily change breaks the pip provider, requiring as a workaround that the administrator create a softlink to pip-python before any pip packages are referenced.

It seems to me that the change could have been accomplished by trying the pip-python alias if the normal exe name failed, rather than assuming that the weird way that one common RPM names pip is a global circumstance.  

Obviously I already discovered the workaround and the problem, so I suppose I'm just posting this in the hopes that it shows up in a Google search for the next guy.  Thanks.

--

M

[Dafydd James]

Thank you! This really helped.

sudo ln -s /usr/bin/pip /usr/bin/python-pip

Or in puppet format:

    file { '/usr/bin/pip-python':

        ensure  => link,

        target  => '/usr/bin/pip',

        require => Package['python-pip']

    }