bitbake/cooker.py: Add support for a bblayers.conf file

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

utility-tasks.bbclass: Remove unneeded base_ prefix since we don't EXPORT these functions any longer

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake/providers: Fix merge error

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

cooker: bail out early if no files to build

This prevents a misleading backtrace:
ERROR: no files to build.
Command execution failed: Traceback (most recent call last):
  File ".../bitbake/build/lib/bb/command.py", line 83, in runAsyncCommand
    self.cooker.updateCache()
  File ".../bitbake/build/lib/bb/cooker.py", line 779, in updateCache
    if not self.parser.parse_next():
  File ".../bitbake/build/lib/bb/cooker.py", line 969, in parse_next
    cooker.bb_cache.sync()
UnboundLocalError: local variable 'cooker' referenced before assignment

(Bitbake rev: 060ef3d957615b7eb1209dc0d01ebeb53f8c4edc)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Chris Larson <clarson@kergoth.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

Fix issues with BBCLASSEXTEND set by anonymous python, and rename some variables to improve code clarity

(Bitbake rev: 3062e96181fe845cfd286990b0216888ddd3d228)

Signed-off-by: Chris Larson <clarson@kergoth.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

Implement ??= operator

??= is a lazy, conditional assignment.  Whereas a ?= immediately assigns to
the variable if the variable has not yet been set, ??= does not apply the
default assignment until the end of the parse.  As a result, the final ??= for
a given variable is used, as opposed to the first as in ?=.

Note that the initial implementation relies upon finalise() to apply the
defaults, so a "bitbake -e" without specifying a recipe will not show the
defaults as set by ??=.  Moving application of the default into getVar adds
too large a performance hit.  We may want to revisit this later.

(Bitbake rev: 74f50fbca194c9c72bd2a540f4b9de458cb08e2d)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

Avoid unnecessary calls to keys() when iterating over dictionaries.

dict objects provide an __iter__ method for the iteration which gives you the
keys, so calling keys directly is unnecessary, and isn't really a best
practice.  The only time you really need to call the keys is if there's a
danger of the dict changing out from underneith you, either due to external
forces or due to modification of the iterable in the loop.  Iterations over
os.environ are apparently subject to such changes, so they must continue to
use keys().

As an aside, also switches a couple spots to using sorted() rather than
creating a temporary list with keys() and sorting that.

(Bitbake rev: 5b6ccb16c6e71e23dac6920cd2df994d67c2587b)

Signed-off-by: Chris Larson <clarson@mvista.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

knotty: tweak failure phrase

..to make copy and paste of the logfile easier.

(Bitbake rev: 446cc0cebd4daff7f849717f4cb89ac1b4c6b755)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Chris Larson <clarson@kergoth.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

Fix ud.basecmd error introduced by the FETCHCMD_git commit

(Bitbake rev: 0bbcbe3548f39ca46c5aa3bf1a8681026e51cbf0)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

Add initial 'repo' fetcher

(Bitbake rev: f68406e864c9837feb56cbec993b620481445cc2)

Signed-off-by: Tom Rini <tom_rini@mentor.com>
Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

Re-getVar BBCLASSEXTEND after finalise()

This ensures that an anonymous python function is able to manipulate the
BBCLASSEXTEND contents, and, therefore, amend.inc files are able to add to it.

(Bitbake rev: c7d038d404afaf4ce3735af5134163759da6f6ef)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

Error early if BBPATH is unset, rather than falling back to share/bitbake

Better to error as early as possible rather than experience strange behavior
resulting from the use of the largely useless stock bitbake.conf/base.bbclass.

(Bitbake rev: 641e6cf3ec3ab4d26929cf4d2a3704ff07eed4d6)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

Kill unused projectdir in bb package.

(Bitbake rev: 79b93e6929c5feeb1ad05bd17f589c69f00b77f6)

Signed-off-by: Chris Larson <clarson@mvista.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

Fix msg usage in the moved utility functions.

(Bitbake rev: 686288444d22091dee66e20ec49b9c53f8c980b7)

Signed-off-by: Chris Larson <clarson@kergoth.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

Move most utility functions from bb into bb.utils.

(Bitbake rev: ff720ec59b30671c951dbf3b96df10ef56b8b505)

Signed-off-by: Chris Larson <clarson@kergoth.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

Move compat imports in the bb package to avoid circular dep issues.

(Bitbake rev: b66c129edc7d78fed9d41b0c634744ec81931b21)

Signed-off-by: Chris Larson <clarson@kergoth.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

Move encodeurl, decodeurl into bb.fetch.

(Bitbake rev: 867d36f9afce2d298874ac7563e5b3852ef04659)

Signed-off-by: Chris Larson <clarson@kergoth.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

Move MalformedUrl, VarExpandError into appropriate modules.

(Bitbake rev: e616483b237dafff7f90ba1c09e9ee7c383a2e47)

Signed-off-by: Chris Larson <clarson@kergoth.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

Add FETCHCMD parameter for git

The Git fetcher currently hardwires the git command to "git". Allow the
path and any additional wrappers to the Git command to be provided via
FETCHCMD functionality, as with some of the other fetchers.

If FETCHCMD_git is not define in bitbake.conf, the fetcher defaults to "git".

(Bitbake rev: f3afb79ecac30d973a3c62ff6baf28d8b7388a24)

Signed-off-by: Malcolm Crossley <malcolm.crossley@ge.com>
Signed-off-by: Martyn Welch <martyn.welch@ge.com>
Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

sanity.bbclass: Add a note about using dpkg-reconfigure

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

base.bbclass: Split up as per the patch in OE.dev by Chris Larson making code more readable and modularised

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

base.bbclass: Really drop siteinfo by default

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

base.bbclass: Don't inherit siteinfo in base.bbclass [based on an OE patch from Chris Larson]

.gitignore: Add handbook generated files to the ignore list

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

xcmsdb: Add libxdmcp to DEPENDS

Signed-off-by: Joshua Lock <josh@linux.intel.com>

claws-mail: Fix building with new automake

Having the same file listed multiple times in Makefile.am breaks with new
Autotools

Signed-off-by: Joshua Lock <josh@linux.intel.com>

poky-fixed-revisions: Bump twitter-glib to a buildable version

Signed-off-by: Joshua Lock <josh@linux.intel.com>

poky-fixed-revisions: Align Clutter revisions with Moblin distro

Primarily to pull in a newer Clutter, which works fixes a build error, but new
versions of software are nice anyway...

Signed-off-by: Joshua Lock <josh@linux.intel.com>

clutter: Fix building version 0.8

Fix building of Clutter with new Automake

Signed-off-by: Joshua Lock <josh@linux.intel.com>

syncevolution: Fix DEPENDS

Add eds-dbus and gnome-keyring

Signed-off-by: Joshua Lock <josh@linux.intel.com>

libsynthesis: Fix DEPENDS

Add sqlite3 and zlib

Signed-off-by: Joshua Lock <josh@linux.intel.com>

luit: add libxdmcp to DEPENDS

Signed-off-by: Joshua Lock <josh@linux.intel.com>

libxscrnsaver: bump version to 1.2.0

Signed-off-by: Joshua Lock <josh@linux.intel.com>

liblbxutil: Bump version and fix dependencies

Signed-off-by: Joshua Lock <josh@linux.intel.com>

handbook: remove OpenedHand contact information

Signed-off-by: Joshua Lock <josh@linux.intel.com>

handbook: Add a section on extending images using packages

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

handbook: Publishing tweaks

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

handbook: Add missing tag

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

handbook/faq: Add entry about building on NTFS

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

task-poky: No longer need to exclude pcmanfm on powerpc

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

pcmanfm: Allow package to build on PowerPC

Signed-off-by: Gary Thomas <gary@mlbassoc.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

zoom2.conf : switch to using the TI integration tree for kernel

Signed-off-by: Graeme Gregory <gg@slimlogic.co.uk>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

linux-omap-zoomsync_2.6.32.bb : add kernel from TI zoom2 integration tree

This is the latest code from dev.omapzoom.org and is also what OE is using
for zoom2 machine.

Signed-off-by: Graeme Gregory <gg@slimlogic.co.uk>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bison: 2.0 -> 2.3

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

handbook: Correction, we're building QEMU x86 not QEMU Arm image

Signed-off-by: Joshua Lock <josh@linux.intel.com>

handbook: update references of stable version number

Signed-off-by: Joshua Lock <josh@linux.intel.com>

handbook: Clean up OpenedHand references

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

handbook: Various tweaks for recent fop/pdf generation

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

poky-doc-tools: Update to work with fop >= 0.95

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

eds-dbus: Add parallel make fixes from Pango

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

task-poky: Disable pcmanfm on powerpc (until confirmed it works)

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

Add i686 SDK machine

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake fetch/__init__.py: Failing urls should raise errors, not fatal exceptions

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

packaged-staging: tweak previous fix to accomodate possible changes to CROSS_DIR

This change, suggested by Chris Larson, ensures that the fix works even if the
definition of CROSS_DIR changes.

Signed-off-by: Joshua Lock <josh@linux.intel.com>

packaged-staging: Fix packagaging of cross packages

packagedstaging_fastpath() was only copying the contents of CROSS_DIR to
PSTAGE_TMPDIR resulting in the folders contents being packaged and then
installed incorrectly at the top level of CROSS_DIR rather than in HOST_ARCH
specific sub directories.

This patch fixes that issue by copying the directory and its contents rather
than just the directory contents.

Signed-off-by: Joshua Lock <josh@linux.intel.com>

packaged-staging: Fix typo, grep not qrep ...

Signed-off-by: Joshua Lock <josh@linux.intel.com>

xf86-video-omapfb: Bump revision.

Signed-off-by: Enric Balletbo i Serra <eballetbo@iseebcn.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

xserver-xf86-config: Do not set default depth to 16 for IGEP v2.

Signed-off-by: Enric Balletbo i Serra <eballetbo@iseebcn.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

xserver-xf86-lite: Fix compilation break for ARM architecture.

This patch syncs the patches used in xserver-xf86-dri-lite and adds
a new patch that fixes a compilation break for ARM architecture.

(http://www.mail-archive.com/xorg-devel@lists.x.org/msg04137.html)

Signed-off-by: Enric Balletbo i Serra <eballetbo@iseebcn.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

gcc: Fix fail to compile since they can't find limits.h include.

With gcc package on my build machine gcc can't find limits.h include.

$ cat x.c

$ gcc -M x.c
In file included from x.c:1:
/usr/include/limits.h:125:26: error: no include path in which to search for limits.h

This patch adds missing includes to gcc package.

Signed-off-by: Enric Balletbo i Serra <eballetbo@iseebcn.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

liboil: 0.3.14 -> 0.3.17

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

eds-dbus: Disable parallel make build problem

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

tasks: Bump to version 0.15

The Owl patch has gone in upstream but requires a different configure switch so
soft assign in the .inc and set appropriately in the 0.15 recipe.

0.16 is out but doesn't build with Poky's Gtk+ version.

Signed-off-by: Joshua Lock <josh@linux.intel.com>

gdb-cross: Allow to build after recent gettext changes

Signed off-by: Gary Thomas <gary@mlbassoc.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

packaged-staging: post-process non-binary files to fix paths in native packages

Extend the existing libtool hack so that native packages (native, cross, sdk)
have all non-binary files scanned and the STAGING_DIR fixed up.

Signed-off-by: Joshua Lock <josh@linux.intel.com>

handbook: tmp/staging was renamed to tmp/sysroots - update references

Signed-off-by: Joshua Lock <josh@linux.intel.com>

preferred-xorg-versions: Bump versions to match available recipes

The PREFERRED_VERSIONS for makedepend and xorg-cf-files where set for versions
which we no longer carry recipes for. Increase them to make BB happy.

Signed-off-by: Joshua Lock <josh@linux.intel.com>

linux-omap: Build fix

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

poky-moblin.inc: Move to meta-moblin

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

glibc: Allow 2.9 to work with binutils 2.20 (patch from OE.dev)

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

poky/moblin: Switch to binutils 2.20

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

binutils: Add 2.20 (patchset from OE.dev)

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

libpcap: Update to 1.0.0 and sync with OE.dev

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

portmap: Sync with OE.dev

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

a52dec: Clean up CFLAGS

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

perl: Update to work with native binary relative path locations

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

relocatable.bbclass: Clean up indentation

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

relocatable.bbclass: Handle symlinks correctly

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

cross.bbclass: Note that relocatable binaries disabled for a reason

Signed-off-by: Joshua Lock <josh@linux.intel.com>

linux: Enable linux-2.6.32 for qemux86 and qemuppc

The qemux86 configuration is simply a copy of the qemux86-defconfig from
linux-rp with make oldconfig and mostly default answers.

Configuration for qemuppc already exists but DEFAULT_PREFERENCE was not set in
the recipe.

Signed-off-by: Joshua Lock <josh@linux.intel.com>

relocatable.bbclass: Enhancements to rpath replacement algorithm

This patch removes the hard coded number of parent directory operators ( /..)
placed into the rpath and instead fully dynamically generates the rpath entries
based on the current rpath.

Theoretically this patch means we can now make cross packages relocatable but
this is *not* enabled as chrpath can only set a new rpath of the same length
or shorter than the existing rpath. Due to the amount of parent directory
jumps we need to encode in the rpath this can easily fail when TMPDIR is in a
short namespace (e.g. /usr/poky).

Signed-off-by: Joshua Lock <josh@linux.intel.com>

bitbake: Revert "providers.py: sort eligible providers by DEFAULT_PREFERENCE"

This breaks preferred providers functionality

This reverts commit ee9afccf33b220a21b74fab279925eeb4771249b.

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

eggdbus: Add missing dbus-glib DEPENDS

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

qmake-naitve: Fix do_install

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [cache] copy instead of deepcopy in sync

We only intend to modify the dict, no need to copy all
elements.

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [parse] In ast, make StatementGroup inherit list, and add a AstNode base class.

(Bitbake rev: 7ea31b2842b45ffe1ca688f55207f8676442a108)

Signed-off-by: Chris Larson <clarson@mvista.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [parse] Use bb.which in resolve_file

(Bitbake rev: 3cd06ed487ee0617892f154cff461379a323a1fb)

Signed-off-by: Chris Larson <clarson@kergoth.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [parser] Make resolve_file only resolve the path

Do not attempt to open the file in the resolve_file method
(a lot like bb.which... maybe bb.which can be used). This way
we don't need to open/close a file which we have already parsed.

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [parser] Cache parsed .inc and .bbclass files for a parse speedup

Have a growing dict with .inc and .bbclass'es. This avoids to reparse
files we have already seen.

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [parser] prepare to cache some .bbcclass and .inc files

Our parser is shit but instead to replace it now we will see
how long we can drive the wave by caching parsed files. This
will not go through the feeder again but we can just reevaluate
the StatementGroup.

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [parser] Remove the "data" from feeder, evaluate after parsing a file

Evaluate the statements after having parsed one file. This is
referred to as "entwirren" and we can remove the direct evaluation
and postpone a bit, in the future we can use a cached copy instead
of parsing the original.

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [parse] Move vars_from_file from bb.parse.BBHandler into bb.parse.

(Bitbake rev: fda0707d772e0964a0185d4ec4d016522f6972f3)

Signed-off-by: Chris Larson <clarson@mvista.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [parse] Don't use relative import of parse_py.

(Bitbake rev: 0a2bb3dd790e3e40867195f14f4e174f98f8a47c)

Signed-off-by: Chris Larson <clarson@mvista.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [parse] Move the last method to the AST...

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [parser] Move more methods to the ast scheme

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [parser] Move one more item to the ast

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [parser] Move methodflags over to the ast

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [parser] Build fix

(Bitbake rev: 9c97696f37499b4d0ec5c034c51e4cf6bc425ba2)

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [parser] Move more statements over the two phase AST

Create the data first, then evaluate on the data dict

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: [parse] Change handling include through the AstNode

Disable the recursively evaluating the statement for now
as it is causing problems.

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>