Fixed:
$ bitbake-prserv-tool import /tmp/1
File "/path/to/lib/bb/parse/__init__.py", line 114, in handle(fn='/tmp/1', data=<bb.data_smart.DataSmart object at 0x2369bd0>, include=True):
return h['handle'](fn, data, include)
> raise ParseError("not a BitBake file", fn)
ParseError: ParseError in /tmp/1: not a BitBake file
But 1.conf or 1.inc works well, check the filename and print proper
error message.
(From OE-Core rev: 273eee7a3614caea17c5bc93b720353641293cf7)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Apparently bb.fetch.URI can't handle this at the moment.
(From OE-Core rev: d3e0a300810251f34932f46daf5263a23846fedd)
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
By default, checkout to latest revision from the machine branch specified by
the user.
(From meta-yocto rev: f79a43406b5b323587415380ecffc87527c64653)
Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fixes the hardcode branch name set to KTYPE, where its value is used as a base branch
when user decides to create a new branch. Tested on x86_64 architecture.
[YOCTO #8630]
(From meta-yocto rev: ab895be90a0cae7dfa77a8aab3b19e5571e7e7bc)
Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
By mistake, the file initially had a wrong extension name, so changing to the
correct one.
(From meta-yocto rev: 32c2278b8fe93429d4cfa097eefccd20157cd3b8)
Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
On linux-yocto-dev or linux-yocto_X.YY bbappend files, the SRC_URI includes
{{=machine}}-standard.scc, which in turn includes {{=machine}}-user-parches.scc,
thus there is no need to include it again on the corresponding bbappend file.
[YOCTO #8486]
(From meta-yocto rev: 11c93b5dd8c651df478d4810e1b6ff6ad9fa57e8)
Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This change removes the file installed_pkgs.txt after it
was used for installation of complementary packages. This
file was causing confusion when left in the WORKDIR after
the build.
(From OE-Core rev: d0f3f3a294d509560bd12b93b26eeec65cfee314)
Signed-off-by: Mariano Lopez <mariano.lopez@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The method _uninstall_unneeded uses the file installed_pkgs.txt,
this file is left after the build and can cause confusion. This
changes allow to get the installed packages using functions of
rootfs instead of the installed_pkgs.txt file. With this change
now is possible to remove the file without breaking anything.
[YOCTO #8444]
(From OE-Core rev: bf935ac16f6175673417dda92a619946b52fac87)
Signed-off-by: Mariano Lopez <mariano.lopez@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If the proxies dictionary has a proxy set to None urllib will throw an exception
instead of not using a proxy (abridged stack):
File: '/home/ross/Yocto/poky/meta/lib/oe/distro_check.py', lineno: 43, function: get_links_from_url
*** 0043: with create_socket(url,d) as sock:
0044: webpage = sock.read()
File: '/home/ross/Yocto/poky/meta/lib/oe/distro_check.py', lineno: 5, function: create_socket
0003:def create_socket(url, d):
0004: import urllib
*** 0005: socket = urllib.urlopen(url, proxies=get_proxies(d))
File: '/usr/lib/python2.7/urllib.py', lineno: 87, function: urlopen
0086: if data is None:
*** 0087: return opener.open(url)
File: '/usr/lib/python2.7/urllib.py', lineno: 203, function: open
0201: else:
0202: proxy = None
*** 0203: name = 'open_' + urltype
0204: self.type = urltype
0205: name = name.replace('-', '_')
Exception: TypeError: cannot concatenate 'str' and 'NoneType' objects
Filter out unset values so that the dictionary only has valid assignments in.
(From OE-Core rev: 2d91290ab5608dd1297d1c26ab807fc4574a8a6b)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The value isn't a string so don't try and expand it.
(From OE-Core rev: ab87d3649c39326938d82d623efafb76905f770d)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Instead of creating two lists of full paths and comparing them which in failure
produces a list of every stamp file (so all tasks, twice), reduce the filename
down to a recipe/task->hash dictionary and compare those, meaning unittest
shows the differences in the dictionaries.
In the future get_files() should be generalised so all tests in this class can
use it, and find a pair of hashes that don't match and run diffsigs on them.
(From OE-Core rev: b612628081b81b50965ae9454df4b2747c6997b2)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
[YOCTO # 8134] This test verifies features introduced in bug 8134.
It requires as resources the files from meta-selftest/files/signing:
For 'gpg --gen-key' the used input was:
key: RSA
key-size: 2048
key-valid: 0
realname: testuser
email: testuser@email.com
comment: nocomment
passphrase: test123
(From OE-Core rev: 6b9d22bfd5414b517a1f0468e1229dfa2294b5fd)
Signed-off-by: Daniel Istrate <daniel.alexandrux.istrate@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The variable name for QEMU_EXTRAOPTIONS is constructed programmatically, so we
need an explicit variable dependency, otherwise changes to it won't cause e.g.
qemuwrapper-cross to be rebuilt.
(From OE-Core rev: 7740f214fffd6278f801899fc5e45f5720cbb544)
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
These need to be based on PACKAGE_ARCH rather than TARGET_ARCH, as we aren't
using overrides for this.
(From OE-Core rev: 46f41df60491990dc41f0514f63b304ac51b67d1)
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
OpenJDK-8 has it's configure script at common/autotools - which will cause
the entire assumption of ${S}/configure is regenerated by autoreconf, intltoolize or alike
fails heavily.
Also - other configure mechanisms can be supported more similar (see how pkgsrc
manages different ones ...)
(From OE-Core rev: fe506eddb0790e37ac1e50f37fa2e32ad81d5493)
Signed-off-by: Jens Rehsack <sno@netbsd.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
TARGET_CPPFLAGS, TARGET_CFLAGS, TARGET_CPPFLAGS and TARGET_LDFLAGS may
differ between MACHINEs. Since they are exported they affect task hashes
even if unused which leads to multiple variants of allarch packages
existing in sstate and bouncing in the sysroot when switching between
MACHINEs.
allarch packages shouldn't be using these variables anyway, so let's
ensure they have a fixed value in order to avoid this problem.
(Compare with 05a70ac30b37cab0952f1b9df501993a9dec70da and
14f4d016fef9d660da1e7e91aec4a0e807de59ab.)
(From OE-Core rev: d08fda21bfb7d264c238af0232a22cdd751f5150)
Signed-off-by: Mike Crowe <mac@mcrowe.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Include stress package in maintainers.inc file
(From meta-yocto rev: a95cb3ef26ead4bd5b2794aa583aba98327e68f0)
Signed-off-by: Maxin B. John <maxin.john@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Hardcoding a full input path with zero flexibility goes against everything the
Yocto Project is about. Rework it to let the user specify the wks base
filename with WKS_FILE and it'll search the layers for the wks file and use
it.
(From OE-Core rev: 8cc7f5229f5447c2183ac319dd52c7ed737ec89b)
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If PACKAGE_DEBUG_SPLIT_STYLE='debug-file-directory', debuglibdir will be
"/usr/lib/debug". Usually 64bit libs should be put under "/usr/lib64".
This often cause an warning, so skip the checking.
(From OE-Core rev: f18a917c1493d4107c7f9db1d9ba19ec368f9b48)
Signed-off-by: Jian Liu <jian.liu@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Immediate expansion of perl_get_libdirs() is too early for
BBCLASSEXTEND. This results in a packaging QA error when building a cpan
recipe as nativesdk. The split debug files do not get picked up by the
dbg package because it looks in the wrong libdir.
The is_target() function remains because it is used elsewhere.
(From OE-Core rev: b32420effcc0435353adc63023d2e4276dd1e0c6)
Signed-off-by: Tyler Hall <tylerwhall@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fix some spelling mistakes in bbclass files
(From OE-Core rev: ed484c06f436eea62c5d0b1a2964f219f3e5cb61)
Signed-off-by: Maxin B. John <maxin.john@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* rootfs_*.bbclass doesn't use this variable anymore, so we can drop it
(From OE-Core rev: bf91547d887aee2893e26e6073e12c540222b422)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The dependency data in BB_TASKDEPDATA is encoded into the sstate checksum
in a much more reliable format. This dependency runs the risk of depending
on the string representation of a dict which is a bad idea. Therefore
remove the dependency.
(From OE-Core rev: 1eefc83e6aeb3cd5501b8e593dda052b1e183cc5)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fixes https://bugzilla.yoctoproject.org/show_bug.cgi?id=8127 which causes
[ 1225.089323] 8189es: Unknown symbol cfg80211_scan_done (err -22)
[ 1225.095916] 8189es: no symbol version for cfg80211_remain_on_channel_expired
when loading external compiled 8189es module.
(From OE-Core rev: afcea61e8eb39234d336c706fdfd4680dea7c060)
Signed-off-by: Jens Rehsack <sno@netbsd.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This doesn't cause any issues right now but it make sense to standardise
on consistently using strings in the data store.
(From OE-Core rev: 99203fbe5ad470ef65cff93cec9d7f332883b5ee)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fixes the following warning sometimes appearing during image builds:
WARNING: The license listed ABC was not in the licenses collected for recipe xyz
The files being looked for here, which runs during do_rootfs,
are written out by the do_populate_lic task for each recipe. However,
there was no explicit dependency between do_rootfs and all of the
do_populate_lic tasks to ensure they had run - only an implicit link via
do_build, so it is possible that sometimes they had not depending on how
the tasks were scheduled. Add an explicit set of dependencies to fix
this.
(From OE-Core rev: ef7dc532e800d9b170246550cbc8703adf624beb)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fixes the following error showing up for layers that aren't a git repo
(or aren't parented by one):
fatal: Not a git repository (or any of the parent directories): .git
This was because we weren't intercepting stderr. We might as well just
use bb.process.run() here which does that and returns stdout and stderr
separately.
(This was a regression that came in with OE-Core revision
3aac11076e).
Fixes [YOCTO #8661].
(From OE-Core rev: f533c1bf4c6edbecc67f9e2c62fd475d64668e86)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The package feed signing code supports the user providing the path to the gpg
binary and an alternative gpg 'home' (usually ~/.gnupg), which are useful for
both deployment and QA purposes.
Factor out the gpg command line construction to a function which can fetch both
of these variables, and also use pipes.quote() to sanitise the arguments when
used in a shell context.
[ YOCTO #8559 ]
(From OE-Core rev: 6daf138822bbbc46960121d3b76b42eaf19e7c0e)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
bbappend files are not included in the archiver during do_ar_recipe.
Find and put them into the tarball.
(From OE-Core rev: 5c84057de5b31c5d6d9abfcca3078bf766a21d88)
Signed-off-by: Jian Liu <jian.liu@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
A previous patch intended to fix and improve the archiver class, and while it
did for gcc packages, with it some issues ended up being hidden, the kernel
tasks taken from kernel.bbclass and kernel-yocto.bbclass specifically expect the
kernel to use work-shared, which either ended up causing issues or wasting time
doing unnecessary work, this patch fixes these issues by performing the right
tasks within the archiver in the right order
[YOCTO #8378]
(From OE-Core rev: d643e43622eb3e43fbb2e21fa33580e2fcdf42be)
Signed-off-by: Alejandro Hernandez <alejandro.hernandez@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
'waf' supports only simple '-j' and fails when parallel make flags
contain extended options like '--load-average'.
Patch uses the method from 'boost.inc' to filter '-j'.
(From OE-Core rev: bc394e1dd229845a315a97704beca43fbb8976ee)
Signed-off-by: Enrico Scholz <enrico.scholz@sigma-chemnitz.de>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This class was extending EXTRA_OECONF only in target builds with
--enable/--disable-installed-tests. However for native builds we don't care
about the test suite and should be explicitly disabling it.
This stops glib-2.0-native trying to build the test suite that we'll never
execute.
(From OE-Core rev: 926a8365b4f7233e5ab5a6b97e8ed53f417c0cfc)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fixes [YOCTO #8696]
Turns out the 'dnf' command is not yet supported for CentOS
as it is for Fedora, I changed the 'dnf' command back to
'yum'. Also, there were some essential packages that needed
to be added to CentOS. Finally, there was a slight
inconsistency in the Fedora list of essential packages and the
ones for supporting Graphics. I had a redundant listing of
one of the packages. I took that out of the Graphics area and
left it only in the essentials area.
(From yocto-docs rev: dde72542ca87bfdd82d713f90938a41c83ac8a4d)
Signed-off-by: Scott Rifenbark <srifenbark@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Since 2.0 release KVM mode does not require VHOST
enablement and a new option was added to support the
old mode. Updated the list of runqemu command options.
(From yocto-docs rev: 6bca626b163ee5e35e04d659579076ab92c4951a)
Signed-off-by: Scott Rifenbark <srifenbark@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Added another link to Josh's porting guide.
(From yocto-docs rev: a1860811628aeddd579f92844f3e6bfcbded2936)
Signed-off-by: Scott Rifenbark <srifenbark@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We have 5.2 and 4.9, we don't really need 4.8 now and it can be moved
out to other layers if anyone still wants/needs it.
(From OE-Core rev: 6f98c39418c60b7c0b25b30983d2e5257158a6a4)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The dependency listed in layer.conf is incorrect, gcc-cross DEPENDS
on ${TARGET_PREFIX}libc-for-gcc, not virtual/libc. These happen to
resolve the same values however they may not always both be built.
The result of this was that gcc-cross gets a different task hash
depending on whether virtual/libc was included in the build.
Specifically "bitbake m4" and "bitbake virtual/kernel" would result
in different task checksums.
The fix is to use the correct dependency name.
[YOCTO #8692]
(From OE-Core rev: 464a4f628f4ec20e2220312e0caa40019edf58ad)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The 'all builds' page was missing the download icon next to the outcome
icon, which allows you to download a build log from the 'all builds' page.
This patch brings it back. It also adds a check in the project builds page
to make sure the download icon only appears if the build generates a cooker
log, since builds that fail at the build request stage do not generate a
cooker log.
(Bitbake rev: 50d7ffa697764d84d96f26a3ae58a9dacde7b31e)
Signed-off-by: Belen Barros Pena <belen.barros.pena@intel.com>
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The project created notifications and the 'Choosea recipe to build' link in
the project page are linking to the software recipes table. Changes them to
link to the image recipes table instead, which is bound to be more useful.
(Bitbake rev: 10e1fbee0d79dbbf15b89ebc155bb2c0092a71f2)
Signed-off-by: Belen Barros Pena <belen.barros.pena@intel.com>
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
(Bitbake rev: c3c7e94cadb3a3a2d8ef33ed456f439c106018c1)
Signed-off-by: Ke Zou <ke.zou@windriver.com>
Signed-off-by: Elliot Smith <elliot.smith@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Our current OVERRIDES handling means we end up caching and checking for
a lot of possible override combinations which turn out to very unlikely.
A typical example is the SRC_URI variable where we have to check if
"URI" is an override. Having spent many hours working in this code, I've
realised all the actual overrides we use are lower case and our standard
variables are mostly uppercase.
This means we could gain quite some speed advantage if we write this
into the code, that overrides only consist of lowercase characters. This
patch shows how simple this is and the resulting speed gains are
significant. This is a significant change but tests show we don't appear
to have any users of capitals in overrides in any OE-Core metadata.
Before "time bitbake -p":
real 2m4.224s
user 7m32.312s
sys 0m7.116s
After "time bitbake -p":
real 1m26.009s
user 5m10.484s
sys 0m4.640s
This check could also be made conditional however I'm not seeing a need
to do that at present.
(Bitbake rev: c9b9443faa76ee7366b1400a56f826f3f9dec1be)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This looks reasonable until you realise self.localpath is a function. Data
expansion of something which isn't a string is the original value so this
code just wastes CPU cycles and makes no sense. Remove it.
(Bitbake rev: 37214ea9bf484998b75dbc1200d53f1afc5257ed)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Trying to expand a variable which isn't a string doesn't make sense.
(Bitbake rev: 62367cca1f1793eb9827406bcdd5980fdeb80a60)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Trying to expand a variable which isn't a string doesn't make sense.
(Bitbake rev: 54d0ddd166a6707b4f8c8535639e3055b783363b)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Instead of putting the DATABASE_URL as part of the command for launching
the bitbake observer process set it as part of environment.
This fixes two issues 1. Where the value isn't quoted and therefore will be
interpreted in the shell and 2. Anyone being able to see the value of
DATABASE_URL in the process tree.
[YOCTO #8669]
(Bitbake rev: 832a8523067606b180c02f0d1544e8a23219bb08)
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
Signed-off-by: Elliot Smith <elliot.smith@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The button required a lot of state maintenance to make sure it
showed up when the project was configured properly, showed correctly
according to the projects known to Toaster, displayed correctly
according to the mode Toaster was in, and was able to be
used to change the current project.
(Bitbake rev: b644514a96f3947ad3f307a26301c064c8ae18f8)
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
Signed-off-by: Elliot Smith <elliot.smith@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Use the project in the template context rather than ajax request to get
the number of ended builds.
(Bitbake rev: 1f6bfa876b17a600fc83c789af8aaa2639a463b2)
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
Signed-off-by: Elliot Smith <elliot.smith@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Robert Yang
Christopher Larson
Leonardo Sandoval
Mariano Lopez
Ross Burton
Richard Purdie
Daniel Istrate
Jens Rehsack
Mike Crowe
Maxin B. John
Jian Liu
Tyler Hall
Martin Jansa
Paul Eggleton
Alejandro Hernandez
Enrico Scholz
Scott Rifenbark
Anibal Limon
Belen Barros Pena
Ke Zou
Elliot Smith
Michael Wood