1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
|
06/11/05: harring: negate added to restriction base, work in package.* and ebuild.package
06/07/05: harring: config.{repo,cache,domain,sync} works fine, new dict type. portage.util.dicts holds the dict variants, including
IndexableSequence. added portage.util.currying. Use it. it rocks.
06/07/05: harring: config instantation of repo works now. config.repo is a bit screwed up, but will be fixed.
06/06/05: harring: ebuild depends works now. config instantiation should work for base configurations, but nothing complex.
cache and location basically, nothing like package.* wrappers (which will be via a config callable)
06/03/05: harring: changelog addition. portage.ebuild.processor work namely, eclass_cache inclusive.
More additions to intro, specifically re: avoiding race condition of gpg verification then using a file with open perms
(short version, abuse preloaded_eclass type trick).
Yanked *ROOTPATH. it's a kludge, a bad one at that. Will break uclibc profile, but it's a kludge to cover up portages shitty path handling.
portage.ebuild.processor.processor.generic_handler is of note, generic waiting loop structure. basically, any ebd interaction after setup
is handled here, with command:callable dict handed in, etc.
ebuild sourcing for metadata should be finished end of day, since I broke it all apart via namespace and killing off old ebuild.py.
ebuild*.sh no longer has /usr/lib/portage/bin/ hardcoded in, if unset and not daemonize ebuild.sh sets it to the usual. ebd requests it
from the processor, which returns portage.const.PORTAGE_BIN_PATH
Broke inherit into two funcs, basically breaking the core of inherit out (the source call/semantics). defaults to old behaviour,
but ebd will override it (look at eclass_cache.get_eclass_* and you'll understand, ellucidates in the intro ebd section)
|
GitWeb
