Bug 528126 – sys-apps/portage-2.2.8-r2 : --newuse should win over --with-bdeps=n

• Product: Portage Development
• Status: RESOLVED CANTFIX

[Description Fab 2014-11-03 14:57:49 UTC]

Created attachment 388426 [details]
emerge --info

On my system, EMERGE_DEFAULT_OPTS is unset. Recently I added « -minimal »
to USE in make.conf, and ran an emerge -avuDN world. Everything seems fine.

>  # emerge -pvuDN world
>
> These are the packages that would be merged, in order:
>
> Calculating dependencies... done!
>
> Total: 0 packages, Size of downloads: 0 kB


But if I do :

>  # emerge -puDN --with-bdeps=y world
>
> These are the packages that would be merged, in order:
>
> Calculating dependencies... done!
> [ebuild     U  ] sys-apps/help2man-1.45.1 [1.43.3]
> [ebuild     U  ] dev-perl/DBI-1.628.0 [1.623.0]
> [ebuild     U  ] dev-libs/libgcrypt-1.5.4-r1 [1.5.4]
> [ebuild     U  ] dev-libs/libxslt-1.1.28-r3 [1.1.28-r1]
> [ebuild  N     ] dev-db/mysql-init-scripts-2.0_pre1-r6 
> [ebuild   R    ] dev-db/mariadb-5.5.40-r1  USE="-minimal*" 
> [ebuild   R    ] virtual/mysql-5.5  USE="-minimal*" 
> [ebuild   R    ] www-servers/lighttpd-1.4.35  USE="-minimal*"

The last three packages will be rebuilt with USE="-minimal".
I think --newuse should win over --with-bdeps=n.

[Comment 1 Zac Medico 2014-11-03 20:42:58 UTC]

(In reply to Fab from comment #0)
> The last three packages will be rebuilt with USE="-minimal".
> I think --newuse should win over --with-bdeps=n.

Those packages will not even be pulled into the dependency graph unless you use --with-bdeps=y, and it's not possible to apply --newuse to packages that have not been pulled into the dependency graph. In fact, packages that are not pulled into the dependency graph are eligible for removal, in order to solve blockers. Furthermore, it's possible to use 'emerge --depclean --with-bdeps=n` to remove those packages. It's not possible for emerge to guess your intentions, therefore you have to explicitly specify --with-bdeps=y if that's the behavior that you want.

[Comment 2 Zac Medico 2014-11-03 20:46:42 UTC]

Note that it's possible that we could add a --with-bdeps=auto which would trigger something similar to the behavior that you have asked for. However, it's not possible to do it within the context of the current --with-bdeps=[y|n] model.

[Comment 3 Fab 2014-11-04 07:32:17 UTC]

Ok. I wanted to be sure this was not a bug before doing anything. I was not aware of this behavior. Then I will add --with-bdeps=y to EMERGE_DEFAULT_OPTS. Thanks for the explanation.