Hi Nick, I noticed this earlier and didn't make sense of it till rac mentioned in #gentoo-dev: <rac> somebody was claiming that portage went through the virtuals list *backwards* as of -r2 And it's true, look at my /var/cache/edb/virtuals entry for virtual/x11 virtual/x11 x11-base/xfree x11-base/kdrive On an emerge -Dup world, it wanted to pull in kdrive as a N (I'd unmerged kdrive a long while ago)
portage devs, I found there are two problems here. This bug does not happen in only one bug. First, reversing virtuals. We use dep_grab_stacked() to make virtual dict now, so virtual is made reversed. Second, using fakedb in dep_zapdep. When it is set --update option, vardbapi==fakedb, so dep_zapdep can't find installed package. I'll fix these problems...
Created attachment 27973 [details, diff] Fix for virtual ordering against 2.0.50-r2 Fixes the ordering of virtuals. Reading of the following files is also affected by this patch but ordering should not affect any of them: use.defaults // Order does not matter make.globals // Variables are not redefined make.defaults // Variables are not redefined package.mask // Order does not matter
Now in cvs. jstubbs, I've modifid the patch a little and commited. Thanks for the patch.
done
*** Bug 49232 has been marked as a duplicate of this bug. ***
*** Bug 49228 has been marked as a duplicate of this bug. ***
*** Bug 49225 has been marked as a duplicate of this bug. ***