| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
Due to an indentation error, atom_arg_map only preserved a single
set reference, preventing slot-operator rebuilds from triggering
for atoms that were also pulled in via another set such as world.
|
| |
|
|
|
|
| |
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
The epoll implementation in qemu-user-1.2.2 appears to generate events
for file descriptors that are no longer registered. Handle it, since it
appears to be harmless (see bug #451326).
|
| |
|
| |
|
|
|
|
|
| |
Since commit f127d25373a42f3200a331b7f2641a1cf0e90b50, --verbose has no
effect.
|
| |
|
|
|
|
|
|
| |
This fixes a case with FEATURE=parallel-install, where a call from
_world_atom to the global event loop could result in reentrace and
lock interference.
|
| |
|
|
|
|
|
| |
to avoid replacing /dev/null character device with a regular file when
using Python 3.4 (http://bugs.python.org/issue17222).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The current output format for listing a chain of dependencies produces
one long flat line that can be hard to read. For example, if you mask
dev-lang/ruby and then try to install dev-ruby/json, you'll see:
The following mask changes are necessary to proceed:
(see "package.unmask" in the portage(5) man page for more details)
#required by dev-ruby/rdoc-3.12.1[ruby_targets_ruby19]
# /etc/portage/package.mask/misc:
=dev-lang/ruby-1.9.3_p385
#required by dev-ruby/rubygems-1.8.24[ruby_targets_ruby18], required by virtual/rubygems-1, required by dev-ruby/racc-1.4.9[-test,ruby_targets_ruby18], required by dev-ruby/rdoc-3.12.1[ruby_targets_ruby18], required by dev-lang/ruby-1.9.3_p385[rdoc], required by dev-ruby/rake-0.9.2.2[ruby_targets_ruby19], required by dev-ruby/json-1.7.7[-test,-doc,ruby_targets_ruby19]
# /etc/portage/package.mask/misc:
=dev-lang/ruby-1.8.7_p371
Tracing your way through that list is not easy. Instead, let's use
newlines and now we get:
The following mask changes are necessary to proceed:
(see "package.unmask" in the portage(5) man page for more details)
# required by dev-ruby/rdoc-3.12.1[ruby_targets_ruby19]
# /etc/portage/package.mask/misc:
=dev-lang/ruby-1.9.3_p385
# required by dev-ruby/rubygems-1.8.24[ruby_targets_ruby18]
# required by virtual/rubygems-1
# required by dev-ruby/racc-1.4.9[-test,ruby_targets_ruby18]
# required by dev-ruby/rdoc-3.12.1[ruby_targets_ruby18]
# required by dev-lang/ruby-1.9.3_p385[rdoc]
# required by dev-ruby/rake-0.9.2.2[ruby_targets_ruby19]
# required by dev-ruby/json-1.7.7[-test,-doc,ruby_targets_ruby19]
# /etc/portage/package.mask/misc:
=dev-lang/ruby-1.8.7_p371
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
|
|
|
|
|
|
| |
Should be no functional change here. The code was indented one too many
times, and the if check is pointless -- the for loop will already skip
if there are no elements in the list.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
|
|
|
|
| |
Shouldn't be any functional changes in here.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
This fixes a case where it could try to pull in an unavailable SLOT.
|
|
|
|
|
|
| |
The "insignificant change" detection from commit
0240b8b754ab5f642c50c863d4bec1dc3636ba29 is only valid when we are
not trying to resolve a slot conflict (see bug #457142).
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
The problem here results from poor handling of the unsatisfied built
slot operator dep inside _add_dep, where it aborts the graph and tries
to backtrack immediately. We really want it to queue a rebuild here,
and continue filling out the graph.
|
|
|
|
|
|
| |
This avoids rebuilding of two packages due to a sub-slot change in the
child package without revision bump. The sub-slot change is considered
insignificant until one of its parent packages needs to be rebuilt.
|
|
|
|
|
|
|
|
| |
This fixes a case where it would return early, before calling
_add_slot_operator_dep, which could prevent slot-operator backtracking
from working properly. This makes SlotChangeWithoutRevBumpTestCase work
properly when we add an undesirable app-arch/libarchive-3.1.1 binary
package.
|
|
|
|
|
|
| |
In the use() IUSE QA check, exempt binary packages for API consumers
like Entropy which do not require a full profile with IUSE_IMPLICIT and
stuff (see bug #456830).
|
| |
|
| |
|
|
|
|
|
|
| |
This avoids weird issues that are triggered when the module is
reloaded (like having classes redefined while instances from the old
classes are still around, triggering TypeError confusion).
|
| |
|
|
|
|
|
|
|
| |
Make child package inherit stable status from the parent package.
This is required in order for USE deps of unstable packages to be
resolved correctly, since otherwise use.stable.{mask,force} settings
of dependencies may conflict (see bug #456342).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If the new profile can't be parsed due to unsupported EAPI, then show a
warning like this:
!!! Your current profile is deprecated and not supported anymore.
!!! Use eselect profile to update your profile.
!!! Please upgrade to the following profile if possible:
default/linux/x86/13.0/desktop
!!! Unable to parse profile: '/usr/portage/profiles/default/linux/x86/13.0/desktop'
!!! ParseError: Profile contains unsupported EAPI '5': '/usr/portage/profiles/eapi-5-files/eapi'
* You must update portage before you can migrate to the above profile.
* In order to update portage, run 'emerge --oneshot portage'.
|
| |
|
|
|
|
|
|
| |
If sub-slot changes without a revbump, trigger a rebuild so that
dependent packages will have the new sub-slot recorded in their
slot-operator deps. Hopefully this will fix bug #456208.
|
| |
|