Hi Maximiliano, thanks for your work on this!
On 29.11.2012 00:46, Maximiliano Curia wrote: > > The script that is used to comment out the interfaces in > /etc/network/interfaces was so buggy that I had to write it completly from > scratch to make it work properly. Nod, the existing script is indeed horrible. > I have added a number of tests, to verify that the script does what it is > supposed to do. I added the cases listed in the reported bugs, and also other > cases taken from different computers I had access to. > > The script comments out the whole block that refers to one iface, not just the > iface line. In the case of mappings, it doesn't comment them out. This could > be changed if it was decided that the desired behaviour was to also comment > them out, I didn't do it because the original script didn't even take them > into account. > > I'm attaching the debdiff that modifies this script and adds the corresponding > tests (executed from debian/rules during the dh_auto_test step). > > I would like to hear opinions about this script from the maintainers or other > interested parties. In the case that I don't get any replies, I would do the > NMU in a couple of days. I will have a closer look this weekend, so expect a more detailed reply then. I just wanted to let you know, that there has been work in the debian-installer (which will land in beta5 or rc1), to write NM system connection configs. This means, if NM is installed during the system installation (e.g. via the desktop task), d-i will write an empty /e/n/i and create configs for /etc/NetworkManager/system-connections/ directly. For the common case it should thus no longer be necessary to run the ifblacklist.sh script. Because of that I have been considering dropping it completely, since not everyone was happy that is has been running unconditionally. So far I refused to drop the script or put it behind a debconf prompt, because that would have meant, basically everybody would see this prompt or have an unmanaged network device. With the updates in d-i, this situation has changed though. The script is still useful for the case, where someone does a minimal installation and later on installs NM / the desktop task. If we are going to keep the script, I'm inclined to now put it behind a debconf prompt, so people can opt-out from the "/e/n/i mangling". I would actually make the default to *not* run the script. That said I'd also like to have further opinions on this, especially from the release team, since this will concern them anyway once I request an unblock. I basically see three options a/ drop the ifblacklist.sh script b/ fix the script and continue to run it unconditionally during the initial installation (what we do today) c/ fix the script and add a debconf prompt, so it is no longer run unconditionally. The default for the debconf prompt would be *not* to run the script. We'd also have to make sure the debconf prompt is not shown in d-i. Michael -- Why is it that all of the instruments seeking intelligent life in the universe are pointed away from Earth?
signature.asc
Description: OpenPGP digital signature
