| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
| |
The portage_uid initialization here must to happend before the
_init_dirs() calls.
|
|
|
|
|
|
|
|
| |
This is handy for prefix installs, since it allows these variables to
be set in make.conf rather than having them hardcoded. Eventually,
the corresponding portage.data constants may be deprecated in favor of
config attributes, since it's conceivable that multiple configurations
with different constants could be used simultaneously.
|
|
|
|
|
|
|
| |
The EPREFIX for the current install is hardcoded here, but access to
this constant should be minimal, in favor of access via the EPREFIX
setting of a config instance (since it's possible to contruct a config
instance with a different EPREFIX).
|
|
|
|
| |
This allows prefix installs to avoid hardcoding archive-dir.
|
| |
|
|
|
|
| |
This reduces the amount of patching needed for prefix installs.
|
| |
|
|
|
|
|
|
|
| |
This adjusts the logic from commit
ff52f9dc31004becb8022e6437088d01917f413c to use FEATURES=force-prefix
instead of USE=prefix. This has the advantage that we don't have to
make any assumptions about the USE=prefix.
|
| |
|
|
|
|
|
|
|
|
| |
This is safe because the prefix flag should be masked in all non-prefix
profiles, and older EAPIs would otherwise be useless with prefix
configurations. This brings compatibility with the prefix branch of
portage, which also supports EPREFIX for all EAPIs (for obvious
reasons).
|
|
|
|
| |
This fixes the case where $PORTDIR doesn't exist prior to sync.
|
| |
|
|
|
|
|
|
|
| |
This fixes an interaction between the code from commits
9ce6da43ab90c4dab97ebf3b8339e5dbc113a0a8 and
cbe44d92ff13b8a22f5b4215b73078ce600c6bf4, so that we don't discard
matches that are identical except for differnces in upper/lower case.
|
|
|
|
|
|
| |
The fixes some cases where emerge.log would show that depclean was
unsuccessful when it was really successful but there was nothing to
remove.
|
|
|
|
|
|
| |
Looping fixes EbuildMetadataPhase failures for ebuilds that produce
more than 4096 bytes of metadata, broken since commit
b432a1b3051d91546649e8f3190675767461d8e8.
|
|
|
|
|
| |
PyPy 1.7 will die due to "libary path not found" if argv[0] does not
contain the full path of the binary.
|
| |
|
| |
|
|
|
|
|
|
| |
Similar to commit b432a1b3051d91546649e8f3190675767461d8e8, don't use
unecessary file objects. It also happens that these changes fix
compatibility issues with PyPy.
|
|
|
|
|
|
|
|
| |
Use a fork to try and PyPy by digesting random data with hashlib
functions. It doesn't look like a bug has been reported upstream for
this yet, so it may or may not be reproducible by others. Anyway, this
allows me to avoid crashing the main PyPy process until I find a real
fix.
|
|
|
|
|
|
| |
There's no need for a file object, and file objects introduce
complexity that can lead to bugs as mentioned in bug 337465 comment 31,
so use os.read() directly on the file descriptor.
|
| |
|
|
|
|
|
| |
It seems saner to check for None, given that _elog_reader_fd is an int,
even though it will probably never be zero.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
This reverts commit 1d4ac47c28706094230cb2c4e6ee1c1c71629aa0.
As discussed on the gentoo-portage-dev mailing list, if we don't have
a compelling reason to sort here then it's just a waste:
http://archives.gentoo.org/gentoo-portage-dev/msg_67b08fcd7a1b832b347caf1b19362cbb.xml
|
|
|
|
|
|
|
| |
This fixes a regression since commit
57cc4e3e8991e7c4394d1dff7698aa62ed2a286b, which make a faulty
assumption that the digraph contained all of the edges contained in
parent_atoms.
|
|
|
|
|
|
|
|
| |
Since commit e544ee13527a0568de2b556510280ed96fc1b3ff, make.defaults
inheritance of variables from make.globals has been broke, since that
commit excluded make.globals from expand_map. This broke settings like
PORTAGE_ELOG_CLASSES="${PORTAGE_ELOG_CLASSES} qa" from the developer
profile, as reported in bug #391323, comment #4.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
This is should never be triggered by portage, but it's possible that
an API consumer might trigger it somehow.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
This will handle invalid message types like the one that triggers the
KeyError in dblink._elog_process for bug #390833. It will also output
the content of the line in order to help diagnose the source of
corruption.
|
|
|
|
|
| |
This is a regression from commit
69d7f169ba2e27cf56746791331c63a019357100, reported in bug #390893.
|
|
|
|
|
| |
There's no need to do this anymore, because we don't restart since
commit d3f704a425a50b5cfa997a25866929b30f1b7d0f.
|
| |
|
|
|
|
|
|
|
| |
Instead, finish the whole job using a copy of the currently running
instance. This allows us to avoid the complexities of emerge --resume,
such as the differences in option handling between different portage
versions, as reported in bug #390819.
|
| |
|