Pkgcore

------------------------

- pmaint regen now supports regenerating binary and install repository
caches.

- pkgcore now tracks and records the originating/source repository
when installing to the vdb.

- new pkg attribute; source_repository. This tracks where a package
originated from- primarily useful for binpkgs and vdb.
pquery --attr source_repository is how to access it from the CLI.

- pkg_config can now be invoked via:
pconfig package <target>

- splitdebug no longer runs if the pkg has been split already.

- arbitrary exceptions during merging/unmerging no longer stop the
merge/unmerge; a traceback is displayed instead.

- added initial profile inspection tool; pinspect profiles.

- pmaint copy arguments have changed; check the help, short version,
it's now sane.

- pkgcore now lives at googlecode; http://pkgcore.googlecode.com/

- large scale conversion of internals to argparse. Saner parsing namely,
although it's still a work in progress to make it pretty.

- man pages and docs in general have been converted to sphinx. Definite
improvement already, but more to come.

- pkgcore observer api's were heavily gutted and split into observer and
outputter. This should enable easier UX integration, while enabling
our next step towards parallelization.


--------------------------

--------------------------

- make use/useq/usev extremely obnoxious towards offending devs who use them
in global scope when they're not supposed to. Pretty much, I'm tired of
pkgcore being broken for being PMS compliant; as such I'm now pointing
users loud and clear at the offenders.

- fix traceback in user profile support (/etc/portage/profiles).


--------------------------

--------------------------

- Log an error, rather than throwing an exception when binpkg cache cannot
be updated. Needs refinement long term, but for average users, this is
preferable.

- loosen up pebuild a bit; choose the max version if slot/repo are all the
same. This allows pebuild dev-util/nano to choose 2.3.1 for example.

- tighten up econf implementation; ctarget/cbuild are now forced as early
arguments to configure to work around some misbehaviours in configure
scripts (broken scripts, but so it goes).

- tighten up ebuild environments variable handling- had a bleed through
of variable 'x' that was breaking mesa builds.

- yet another src_install fix for EAPI=4; this time ensuring the default
function is available.

- we now run bashrcs (profile and user) every phase to match portage
behaviour. If folks desire it, a patch making that optional would be
welcome.

- add support for /etc/portage/package.env and /etc/portage/env/. Note
that we only allow settings there to affect the bash environment- trying
to adjust FEATURES from those files isn't on the intended support list.

- use ${LIBDIR_${ABI}} for ccache/distcc pathways; gentoo bug 355283.

- profile interpretation of make.defaults now has access to variables
defined by its parents, per PMS.


--------------------------

--------------------------

- intercept and suppress exceptions from triggers unless the trigger
explicitly disables it.

- work around libmagic python bindings sucking and not always being
able to be used.

- fix 'default' support for src_install for EAPI=4.


--------------------------

--------------------------

- support for /etc/portage/make.profile; Please Do Not Use it, while
pkgcore is forced to support it, usage of it breaks most tools and is
bluntly lock-in (no reason to move it- it's the same, been in the same
place for a decade now). Duly warned.

- misc env/bug fixes for EMERGE_FROM to ensure compatibility.

- deploy eselect support via pinspect portageq

- added man page for pinspect

- added pmaint env-update

- expose /usr/local/* through PATH for ebuilds.


--------------------------

--------------------------

- for EAPI<4, expose MERGE_TYPE info via EMERGE_FROM; do this for compatibility
with non-spec compliant ebuilds, and eclasses like linux-mod. This restores
in particular, binpkg support for kernel packages. Thanks to Brian De Wolf
for info leading to tracking this down.

- add support for stacking /etc/portage/make.conf on top of /etc/make.conf.

- add incrementalism between make.globals and make.conf to match changes
in portage configuration parsing. This fixes the common "I tried pkgcore
and everything was license masked". Breakage there owes to portage
changing make.globals; can't do much about it unfortunately. Thanks to
Brian De Wolf for info leading to tracking this down.

- prefer 0755 permissions for binpkg package dir.

- pinspect pkgset learned --all option, to display all pkgsets it knows.


--------------------------