| Age | Commit message (Collapse) | Author | Files |
|---|
|
|
|
|
|
is not present
This is necessary due to the new RDEPENDS handling. I don't feel 100% comfortable with it,
but until we rework the complete staging-shouldn't-contain-dependency-information, this is
our way.
|
|
|
|
|
|
and e6a2ccb39d3b001fdcc1071ba1860678a1fe8647
|
|
-We do not have rpath on Darwin, do not use it
for the native builds
|
|
Set BB_MIN_VERSION here
classes/sanity.bbclass:
-Run the sanitycheck exactly once (on BuildStart)
-Check for common utilities/executables as well
|
|
|
|
|
|
|
|
e17 packages, update preferred e versions
|
|
|
|
|
|
misconfigurations and gives helpful information about fixing them. The checks can be expanded over time. It can be disabled by following the slightly cryptic instructions in sanity.conf.
|
|
-Exchange <--- and ---> to fix the tinderbox
view
-Add a new line to fix the tinderbox display
for multiple providers.
|
|
Use the MultipleProvider and NoProvider Event to create
reports to the TinderBox.
The message format can be improved now we at least know
if a build failed due a missing provider
|
|
Send the name of the to be build packages. Now
we can see what was intended to be build from
within the log
|
|
using qemu. based on a patch originally written by Phil Blundell. fixes hh.org Bug #1385 (localedef OOM issue).
- running localedef (in qemu...) for every locale takes a while. set
BINARY_LOCALE_ARCHES = "" to avoid the new behaviour.
|
|
PARALLEL_INSTALL_MODULES, courtesy Raymond Danks
|
|
|
|
-Disable sending of the environment as this somehow
refuses to work from within a screen - dunno why
-The lines of the log messages are ended with a new line
no need to add another empty line.
|
|
-Have a none empty log messages when the build is
completed otherwise the log is not transfered
to the server.
Now the build turns green :)
|
|
Be less verbose do not print the results for now.
A proper fix/configurable one is pending.
|
|
Redo the changes... they were magically lost.
readlines is a sequence, we join them (we might
need to remove the \n)
|
|
|
|
-Send the right messages to the TinderBox
-Read the log from the logfiles and send it to the tinderbox
-Add the wish for a more sane and more featurefill reporting
(e.g. less data, less often)
|
|
-Save the assigned machine id to a file
and restore it when needed. BitBake does
not have a 'persistent' cache one could use
across files...
-Post the configuration directly to the server
|
|
Begin of the rewrite to support the tbox3 protocol
|
|
|
|
|
|
-Add the skeleton for the QA bbclass
|
|
package.bbclass now tracks any package renaming performed using PKG_*.
When a package is built, the runtime variables are translated using
this cached information which is stored in staging.
From now on, only use package names are valid in runtime variables like
RDEPENDS, RRECOMMENDS and RSUGGESTS. eg. fontconfig-dev is correct,
libfontconfig-dev would be wrong.
As debian.bbclass can't determine package renaming until after a
package's contents is known, the class has to set BUILD_ALL_DEPS = "1".
This also means OE is no longer forced into one package renaming scheme.
|
|
specific packages.
|
|
|
|
|
|
patches courtesy Martin Dietze <dietze@4g-systems.com>
Martin: I'm not sure whether you really need a dedicated class for nylon image,
you may want to think about that. the patch against the mtx-1 machine configuration
didn't apply, please handle this on your own after you got commit access.
|
|
|
|
|
|
IGNORE_STRIP_ERRORS
|
|
- package.bbclass now uses file-native and cross strip to reliably strip
unstripped executables and check the return code. For the moment a
failure here doesn't cause the build to fail but does output a failure
message direct to the controlling terminal (this is temporary). This
behaviour can be changed by forcing IGNORE_STRIP_ERRORS = ""
The change works by using a new runstrip shell function for every
potentially strippable file. Since this calls native 'file' there is
a dependency on all inheritors of package.bbclass for file-native and
since this would break native builds (because of the cycle in
file-native and it's own native depends) native.bbclass cancels this
dependency (system file should be fine for native packages and, anyway
native currently doesn't package.)
|
|
|
|
failure.
|
|
|
|
- define TARGET_KERNEL_CC_ARCH to define arch specific compiler options
in machine.conf. The default for the kernel is "" (empty), although
it should probably be TARGET_CC_ARCH in the future. The default for
modules is TARGET_CC_ARCH
|
|
IGNORE_STRIP_ERRORS to get the old behaviour back. Part of a fix for hh.org Bug #1469
|
|
to access GPIO registers. make sure to bump PR on all affected .bbs. thanks to Phil for spotting.
|
|
doesn't exist for the unslung 2.4 kernel).
|
|
and f1adf0c8840c57a416d7da7085aa694f11be4c5c
|
|
all RDEPENDS
|
