| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
| |
It isn't used externally anymore, since SchedulerInterface is used
directly in those places now. Many of the self.sched_iface references
updated here, it's more appropriate to use self._event_loop.
|
|
|
|
| |
See bug #437516 for example.
|
|
|
|
| |
This will fix bug #437546.
|
|
|
|
|
|
| |
SchedulerInterface suffices for all of these cases.
EventLoop(main=False) is used for thread safety where
API consumers may be using threads.
|
| |
|
|
|
|
|
| |
The PollScheduler class suffices here, if we just add a small amount of
logic to calculate the _background attribute from the emerge opts.
|
|
|
|
|
|
| |
- Print different die() message for invalid atoms.
- Mention ebuild-ipc instead of portageq in other die() messages when
ebuild-ipc is actually used.
|
| |
|
| |
|
| |
|
|
|
|
| |
This ensures that errors are logged before new jobs are scheduled.
|
|
|
|
|
| |
EventLoop suffices for all of these cases. EventLoop(main=False) is
used for thread safety where API consumers may be using threads.
|
|
|
|
|
|
|
| |
These methods were aliases for the EventLoop io_add_watch and
source_remove methods. Migrating to the EventLoop method names allows
an EventLoop instance to substitute for a PollScheduler inside
subclasses of AbstractPollTask.
|
|
|
|
| |
This is variable is only needed by the Scheduler class.
|
|
|
|
|
|
| |
It's not allowed for AsyncScheduler to inherit from two classes that
define __slots__, so it useless to define __slots__ for AsyncScheduler
and subclasses.
|
|
|
|
| |
This allows the QueueScheduler class to be eliminated.
|
|
|
|
|
| |
Emulate the sleep command, in order to ensure a consistent return code
when it is killed by SIGTERM (see bug #437180).
|
| |
|
| |
|
| |
|
|
|
|
|
| |
If the existing Manifest already has the correct content, but it is
not signed, then sign it if appropriate.
|
|
|
|
| |
Triggers cleanup and exit listeners if there's nothing left to do.
|
| |
|
|
|
|
|
| |
The _terminate_tasks method is always supposed to be called via
_termination_check.
|
| |
|
|
|
|
|
| |
Also add missing __slots__ to ForkProcess. TODO: Share code
between ForkProcess and MergeProcess.
|
|
|
|
|
|
|
|
|
|
| |
Update manifest files, and sign them if signing is enabled. This
supports parallelization if enabled via the --jobs option. The
--thin-manifests and --sign-manifests options may be used to manually
override layout.conf settings. There's also a new --strict-manifests
option that may be used to manually override the "strict" FEATURES
setting, a --gpg-key option to override PORTAGE_GPG_KEY, and a
--gpg-dir option to override PORTAGE_GPG_DIR.
|
| |
|
|
|
|
|
|
| |
This fixes some strange behavior triggered during fetch, which is only
observable with Python 2.6 since it doesn't have
collections.OrderedDict.
|
|
|
|
|
|
| |
This fixes warnings for prefix users (see commits
1c3c3b802014ac6356af09d149aaaffce6c14dc3 and
6983b1a1dd21f931ba751b727bd1a3c460383840).
|
|
|
|
|
|
|
|
|
|
|
|
| |
This partially reverts commit 314791e319e3ac7ce75f3a6f9ae9cc5c2b837df6,
in order to cover possible cases where it may not actually be
redundant, like if ebuild(1) is used to call pkg_setup as root, and
then a different user calls src_unpack. We use PORTAGE_GRPNAME if set,
in order to avoid triggering irrelevant warnings for unprivileged
prefix users (see chown and chgrp wrappers in commit
45bedf9cdc10fafd94858f67d62b36d35dd99544). This mirrors the approach
taken in commit 1c3c3b802014ac6356af09d149aaaffce6c14dc3 of the prefix
branch.
|
|
|
|
|
|
| |
These calls trigger warnings in prefix mode, since commit
45bedf9cdc10fafd94858f67d62b36d35dd99544. They are redundant, since the
permissions are handled the _post_phase_userpriv_perms function.
|
|
|
|
|
|
|
| |
For prefix, there is typically only a single unprivileged user who owns
everthing that is built and installed, and in this case it's desirable
to ignore chown and chrgrp failures. So, move the logic from fowners
into chown and chgrp wrappers, as discussed in bug #433453.
|
|
|
|
| |
See http://forums.gentoo.org/viewtopic-t-938022.html for example.
|
| |
|
|
|
|
|
|
| |
This probably makes no difference in practice, since it only affects
the "depend" phase. Note that we're allowed to use EAPI conditionals
here, since we parse the EAPI on the python side and pass it to bash.
|
| |
|
|
|
|
| |
This broke in commit ab46499322311c1faa710c63d0a5339e49a9061a.
|
|
|
|
| |
This is needed since commit ab46499322311c1faa710c63d0a5339e49a9061a.
|
|
|
|
|
| |
If /etc/portage/profile/eapi does not exist, then support extended atom
syntax. This will fix bug #436330.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
This fixes a case where EventLoop could consume 100% CPU while waiting
for a child process. It also fixes timeout calculations in python 2.x,
where it was using integer division instead of float.
|