-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 On 16/08/13 10:57 AM, Todd Goodman wrote: > * Rich Freeman <ri...@gentoo.org> [130816 10:43]: >> On Fri, Aug 16, 2013 at 10:09 AM, Markos Chandras >> <hwoar...@gentoo.org> wrote: >>> The package is now masked (openrc-0.12) because quite a few >>> people lost their net configs >>> >>> So yep, ~arch being *this* broken is not so nice >> >> And hence the value of having a group of volunteer guinea pigs >> (anybody running ~arch) is demonstrated. That said, masking big >> changes and calling for volunteers among the volunteers doesn't >> hurt. >> >> Seems like we need to be more careful with code that runs outside >> the sandbox. Config protection is nice, but it is useless when >> code runs outside the sandbox. >> >> Rich > > As one of those volunteer guinea pigs it all worked fine with the > four ~x86 and three ~amd64 machines I've upgraded to > openrc-0.12:0. > > They vary in when they were installed from 2005 up to a couple > months ago and are generally updated daily. > > All ~x86 are "servers" (though most have X, KDE, and Gnome > installed, they're only accessed remotely.) > > Two of the ~amd64 machines are "desktops" (though they both run > services as "servers.") > > If I can help narrow anything down further I'm happy to help. Or > to test anything. >
For everyone's information -- The conf.d/net removal on upgrade is a packaging issue, which could not have been tested prior to openrc-0.12.ebuild hitting the tree. There are details in https://bugs.gentoo.org/show_bug.cgi?id=481336 if anyone's interested in why it's happening. I've fixed the 0.12.ebuild in the tree now. It's a hack but it seems to be the best possible solution. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.20 (GNU/Linux) iF4EAREIAAYFAlIOT+QACgkQ2ugaI38ACPB5cAD+KZaB/IOhTOQB90L5JEwPnBcO eJzbFHOqtxeJAQ/i6pgBAKukByT2wFolArwBoNxjo6e+D+uVEw+Rct2KPL3cXM7t =NhhK -----END PGP SIGNATURE-----
