| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
for when using pipes
|
| |
|
| |
|
|
|
|
| |
from there, just defining package classes and metadata pulls; the mutable aspect comes later
|
|
|
|
| |
forced, but still must be tracked for rewinding the stack)
|
|
|
|
| |
2) made use/slot dep restriction generation live (so it's now supported in the atom spec)
|
|
|
|
|
|
| |
2) full use deps implemented, and slot deps.
3) reverse use dep lookup, effectively "yo, turn off depends x please" resulting in figuring out how, and configuring the package appropriately
4) bug in obj caching for tracking when forced to regenerate an obj (package.conditionals)
|
|
|
|
|
|
| |
the new seperation brought to light
2) corrected negated use flags so they properly come through as negated, not -use_flag
|
|
|
|
| |
imports/code for collapsed for the tweaked restriction design
|
|
|
|
|
|
|
|
|
| |
bugs probably exist in force_(Fail,True), although containmentmatch appears to work (use deps are working now)
packages is package level restrictions
values is value level restrictions; package restrictions gank the attributes, and hand them off to value restrictions (effectively)
collapsed is package level.
boolean is a base of boolean logic classes, properly support the full potential combinations NAND and OR have (that sucked).
same with ContainmentMatch
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
package trees, creating one from an unconfigured tree during
domain instantion.
Package returned are mutable within limits of use restrictions (arch included); all depset (for ebuilds) returned are flattened, no use conditionals exposed.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
initial implementation, probably will be overhauled a bit, but right now is usable for screwing with the default repository.match
|
|
|
|
| |
added class configured=True attribute, in preparation for commit of mutable configurable wrapper for portage.ebuild.repository (use deps/slot deps, yay)
|
|
|
|
|
|
| |
throwing of said exceptions), dropped __str__ (so people now see the underlying portage.restrictions.restrictionSet.AndRestrictionSet.__str__ output, which is better for hashing)
and a __cmp__ function, which is of questionable use.
|
|
|
|
|
|
|
| |
1) renamed base repository tree to UnconfiguredEbuildTree (since it produces non-configured packages).
2) portage.util.dicts => portage.util.mappings changes
documentation additions also.
Threw out ConfiguredEbuild from ebuild_package, added start of a configured ebuild tree/wrapper to ebuild_repository.
|
| |
|
|
|
|
|
|
| |
and updated portage.ebuild.conditionals
pydoc documentation also...
|
|
|
|
|
|
| |
addition of __all__ where I've decided (collapsed and restrictionset at this point)
addition of cmatch method, used for dealing with mutable configuration wrappers (ongoing)
yanked SubStringMatch, since it's covered by ContainmentMatch (fixed that beast too, since first implementation of it sucked).
|
|
|
|
|
| |
cparser holds the case sensitive configparser (broke it out of __init__)
documentation updates, and addition of most filters for domain, including license filters, p.mask, and keywords.
|
|
|
|
| |
aside from that, removed a lot of older portage funcs from fs.py, and added pydoc documentation
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
to do a regen.
|
| |
|
| |
|
|
|
|
|
| |
s:pref_curry:pre_curry: in central
hey look, domain has code now, and domain-instance.repos (which is visibility wrapped, p.mask + unmask atm only)
|
|
|
|
|
|
| |
s:pref_curry:pre_curry: for proccessor
ebuild_package had USE added as an attribute...
and conditionals got cleanup + __str__
|
| |
|