| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
| |
Signed-off-by: Arthur Zamarin <arthurzam@gentoo.org>
|
| | |
|
| |
|
|
|
|
| |
The bash side calls filter_env via IPC now so this script is mostly
pointless. It could be much more useful if it was reimplemented as a
native bash plugin that outputted a filtered env dump directly.
|
| |
|
|
| |
Fixes #45
|
| | |
|
| |
|
|
|
|
|
| |
This currently supports distfiles, binpkgs, and tmpfiles. Config file
support and more tests coming later.
Fixes #163.
|
| |
|
|
|
| |
The setup_requires attribute for setuptools should work well enough to
pull in snakeoil before the setup script is run.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
This rids us of the pwrapper/pwrapper_installed linking method. Now the
initialization code lives in the main pkgcore.scripts module which
handles sys.path munging for repos and tarballs as well as being used as
a target for installed scripts that are generated during the build
process.
We could trivially use entry_points from setuptools to generate and
install the scripts as well, but that appears to add a relatively
significant amount of extra runtime for scripts executing under a second
due to all the extra modules pkg_resources imports. Also, we probably
don't need the extra complexity and layers of abstraction anyway.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
During installation, setup.py will now write out a pkgcore._const file
holding the pathways that were specified during the install- primarily pathways
to get at certain data files, or ebd directories.
In the absence of this file, it's assumed the code is being ran from a non
installed state- git checkout or release tarball- and defaults to finding that
content within the repository.
Additionally, pkgcore.const grew a _GET_CONST function; this function pulls
in content from pkgcore._const if it exists, allows for interpolation against
other const values, and supports pulling the variable out of the environment
(if the const allows it); any env access is prefixed with PKGCORE_OVERRIDE_.
setup.py test uses this environmental support to override the data path used;
very little else should need that sort of support, but it's there.
Note: this requires snakeoil 27a76bc or later due to reliance on a new
snakeoil.mappings.ProxiedAttrs class.
|
| |
|
|
|
|
| |
Additionally, move PKGCORE_REPO_PATH environmental setting to prior
to importation. No code should be sensitive to it during import, but
it's possible; thus the change.
|
| | |
|
| |
|
|
|
|
|
| |
We don't need to duplicate these for every version of python we install
for especially once py3 support is re-added. Note that there is one
python script in the bash directory for dohtml, but it's always had
a generic shebang anyway.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
Matches pplugincache naming and makes it easier to autogen a few more
docs.
|
| |
|
|
|
|
| |
This follows the sys.path docs that state the first item of sys.path
should be the directory containing the script used to invoke the
interpreter.
|
| |
|
|
|
|
| |
We should try to keep these as similar as possible so it's clear how
they differ (currently just munging sys.path and adding an env variable
for the git repo wrapper script).
|
| |
|
|
|
| |
All scripts now use argparse so we don't need to keep the old fallbacks
around.
|
| |
|
|
|
|
|
|
| |
Instead of using python shim scripts for pinspect and filter-env for
calls from the bash side we use the versions found in PATH. This is done
for non-system installed pkgcore by adding the bin directory (which
already contains the relevant python scripts) of the git repo or
unpacked tarball to PATH.
|
| | |
|
| | |
|
| |
|
|
|
| |
Add python version, and match pwrapper (used for development) to the
production code.
|
| | |
|
| |
|
|
| |
filter-env) installed.
|
| |
|
|
| |
snakeoil.klass code. at this point, pkgcore works under py3.1 w/out issue
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
out into snakeoil, etc.
|
| | |
|
| | |
|
| |
|
|
| |
util.commandline'ifying lot of my old scripts, lot of pquery output purdying up.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
Shifted the fds to the end of the range to avoid any idiot scripts writing to an fd they shouldn't be.
|
| |
|
|
| |
restriction generation
|
| | |
|
| |
|
|
| |
Not perfect, but tiz a start.
|
| | |
|
| | |
|
| |
|
|
|
|
| |
1) move install/compile vars to be defined by dyn_compile.
2) fix single var/func filtering so they don't dump a newline
3) disable attribute dumping during load_environ, use what's in the env instead
|
Arthur Zamarin
Tim Harder
Brian Harring
Tim Harder
GitWeb