Created attachment 437248 [details, diff] Patch to create the required 'mail' group during building During compilation mutt executes chown root:mail $SOMETHING When the 'mail' group doesn't exist, this fails. The attached patch creates the required group during pkg_setup()
mutt unconditionally depends on net-mail/mailbase and mailbase creates mail group. You are most likely doing something wrong. Please attach output of `emerge --info' and complete build.log where the problem can be seen.
(In reply to Coacher from comment #1) > Please attach output of `emerge --info' and complete build.log where the > problem can be seen. I think I know what happened. I set this system up through ROOT= (gentoo-embedded). net-mail/mailbase probably got depended in when I installed sendmail, but pkg_setup() won't be executed in this case. How would this be done properly?
Created attachment 437406 [details] Just for reference: emerge --info as requested.
(In reply to Dirk Tilger from comment #2) > I set this system up through ROOT= (gentoo-embedded). > How would this be done properly? Sorry, I don't know the right way to do cross compilation in Gentoo.
I guess mailbase should also be in RDEPEND. If you move the mailbase dep to CDEPEND, does it properly pull it into your ROOT then?
(In reply to Fabian Groffen from comment #5) > I guess mailbase should also be in RDEPEND. If you move the mailbase dep to > CDEPEND, does it properly pull it into your ROOT then? mailbase was actually installed on that box, so I think it's not a problem of dependencies. However, the useradd call would have to be executed within a chroot environment if not(ROOT=/) and as far as I could see in user.eclass that doesn't happen. In the past I've had to rerun the pkg_setup() phase for packages like apache so they create their user accounts and set up the system (with CTRL+C once they are in the configuration phase), I was just wondering if there isn't a better way to do this.
That's interesting indeed, but unrelated to Mutt. Perhaps we could turn this into a user.eclass bug instead?
(In reply to Fabian Groffen from comment #7) > That's interesting indeed, but unrelated to Mutt. Perhaps we could turn > this into a user.eclass bug instead? https://bugs.gentoo.org/show_bug.cgi?id=541406
(In reply to Fabian Groffen from comment #7) > That's interesting indeed, but unrelated to Mutt. Perhaps we could turn > this into a user.eclass bug instead? Agreed, let's close this bug.