| 1 |
On Fri, 15 Mar 2013 13:47:49 +0800, Mark David Dumlao wrote: |
| 2 |
|
| 3 |
> > There's no need to rebuild everything, and those other flags make no |
| 4 |
> > sense when using -e. Generally you only need |
| 5 |
> > |
| 6 |
> > emerge -uaD --changed-use @world |
| 7 |
> > |
| 8 |
> |
| 9 |
> I know that, in general principle. But it's a test environment. I'd |
| 10 |
> assume stricter standards of "purity" there than elsewhere. simply |
| 11 |
> going by changed-use can break some library dependencies. We need to |
| 12 |
> use depclean to remove build deps junk after the emptytree, and we're |
| 13 |
> revdep-rebuilding twice in case the depclean borked something. (To be |
| 14 |
> really strict, revdep-rebuild should be repeated until it stops |
| 15 |
> building things...) |
| 16 |
|
| 17 |
portage should handle that itself nowadays, but it doesn't hurt to run |
| 18 |
revdep-rebuild to be sure. You could use -N instead of --changed-use but I |
| 19 |
still think -e is unnecessary. |
| 20 |
> |
| 21 |
> Heck in some setups empty-tree will simply fail thanks to circular |
| 22 |
> deps of the global use flags and you'll need manual intervention to |
| 23 |
> bootstrap a package with less USE... |
| 24 |
|
| 25 |
And that's a good reason to not use -e. If you do use -e, none of the |
| 26 |
other options make any sense, -u -D and -N are meaningless if the system |
| 27 |
thinks nothing is installed and there's no point in using -t without -a |
| 28 |
or -p, and with -e it would generate so much output I'm not sure many |
| 29 |
people would bother reading it all. |
| 30 |
|
| 31 |
|
| 32 |
-- |
| 33 |
Neil Bothwick |
| 34 |
|
| 35 |
Cross-country skiing is great in small countries. |
Archives