On 19-Oct-2015, Piotr Ożarowski wrote: > [Ben Finney, 2015-10-18] > > > I have experienced the same error. A deliberately-created > > > symlink causes ‘dh_python3’ to crash when it attempts to > > > ‘os.rmdir’ the symlink > > (symlinks to dirs are bad, very bad for dpkg, so dh_python* should > fix them or fail with a better message)
How are they bad? Searching for “dpkg” and “symlink” doesn't help me to know what you're referring to. Can you point us to some official Debian packaging documentation to explain what that means? The symlink to a directory is deliberately created in the packaging, using ‘dh_link’. I would expect that to be a normal packaging product, not problematic for dpkg. The upstream code expects the presence of a directory in among the Python modules, but in the Debian package, that directory is located under ‘/usr/share/$PACKAGE/…’ instead. So a symlink is put in its place. I would expect dh-python to install the symlink as a symlink to the real directory, and IIUC that is what this bug report is requesting. Are you suggesting some other behaviour to resolve this? Can we know what you propose, because I'm concerned it will not actually resolve the reported bug. -- \ “The sun never sets on the British Empire. But it rises every | `\ morning. The sky must get awfully crowded.” —Steven Wright | _o__) | Ben Finney <b...@benfinney.id.au>
signature.asc
Description: PGP signature
