[oe] gitver problems

[Cliff Brake]

As I continue to ponder this, it seem it makes sense to simple include
the bitbake recipe directly in the application source.  This makes it
very easy and consistent to set the source directory 'S'.  The big
advantage with this is we easily know the location of the source tree
-- its just the same place as the bitbake recipe.

============================
# recipe to build the autotools-demo in the working tree

inherit srctree autotools gitver

PV = "${GITVER}"
S = "${FILE_DIRNAME}"
============================

Now, anyone can add this application to their recipe tree very simply by:

cd <oe>/recipes
git clone git://git.bec-systems.com/autotools-demo
cd -
bitbake autotools-demo

Now, I no longer see the below errors, so that is good.  However, I
see following problem:

cbrake at happy:/scratch/oe/oe-build/oe$ bitbake autotools-demo
NOTE: Handling BitBake files: \ (8385/8385) [100 %]
Parsing of 8385 .bb files complete (7917 cached, 468 parsed). 8306
targets, 333 skipped, 1 masked, 0 errors.
NOTE: Resolving any missing task queue dependencies
NOTE: Preparing runqueue
NOTE: Executing runqueue
NOTE: Running task 723 of 733 (ID: 0,
/scratch/oe/oe-build/oe/openembedded.custom/recipes/autotools-demo/autotools-demo.bb,
do_populate_sysroot)
NOTE: Running task 724 of 733 (ID: 6,
/scratch/oe/oe-build/oe/openembedded.custom/recipes/autotools-demo/autotools-demo.bb,
do_distribute_sources)
NOTE: package autotools-demo-1.0-1-ga33ef01-r0: task
do_populate_sysroot: Started
NOTE: autotools-demo-1.0-1-ga33ef01-r0: executing task do_configure
NOTE: package autotools-demo-1.0-1-ga33ef01-r0: task do_configure: Started
NOTE: package autotools-demo-1.0-1-ga33ef01-r0: task
do_distribute_sources: Started
NOTE: package autotools-demo-1.0-1-ga33ef01-r0: task
do_distribute_sources: Succeeded
NOTE: package autotools-demo-1.0-1-ga33ef01-r0: task do_configure: Succeeded
NOTE: autotools-demo-1.0-1-ga33ef01-r0: executing task do_qa_configure
NOTE: package autotools-demo-1.0-1-ga33ef01-r0: task do_qa_configure: Started
NOTE: package autotools-demo-1.0-1-ga33ef01-r0: task do_qa_configure: Succeeded
NOTE: autotools-demo-1.0-1-ga33ef01-r0: executing task do_compile
NOTE: package autotools-demo-1.0-1-ga33ef01-r0: task do_compile: Started
NOTE: package autotools-demo-1.0-1-ga33ef01-r0: task do_compile: Succeeded
NOTE: autotools-demo-1.0-1-ga33ef01-r0: executing task do_install
NOTE: package autotools-demo-1.0-1-ga33ef01-r0: task do_install: Started
NOTE: package autotools-demo-1.0-1-ga33ef01-r0: task do_install: Succeeded
FATAL: Unable to start to 'None' UI due to exception: 23634.
ERROR: TaskFailed event exception, aborting
ERROR: Build of
/scratch/oe/oe-build/oe/openembedded.custom/recipes/autotools-demo/autotools-demo.bb
do_qa_staging failed
ERROR: TaskFailed event exception, aborting
ERROR: Build of
/scratch/oe/oe-build/oe/openembedded.custom/recipes/autotools-demo/autotools-demo.bb
do_package failed

If anyone is interested in reproducing this problem, it should be very
simple by running the above instructions.

This concept seems fairly useful to me in that it allows device
developers to easily add their custom applications to an OE build, and
everything is on one place.  I think the existing separation of recipe
and work tree makes sense for canned stuff that OE provides, but
device developers need an easier way to do iterative target
development, and I think a working srctree/gitver gives us that.  The
git submodule mechanism is very exciting in that it gives us a
standard way to lock down versions of various submodules without
reverting to manually specifying hash values in recipes, etc.

Thanks,
Cliff

On Mon, May 31, 2010 at 5:12 PM, Cliff Brake <cliff.brake at gmail.com> wrote:
> Hello,
>
> I'm trying to put together some examples of using srctree/gitver, and
> running into a few issues.
>
> My example is in a bitbake overlay:
>
> git://git.bec-systems.com/openembedded.custom
> http://cgit.bec-systems.com/cgit.cgi/openembedded.custom/
>
> This overlay includes a submodule for:
>
> git://git.bec-systems.com/openembedded.custom
> http://cgit.bec-systems.com/cgit.cgi/autotools-demo/
>
> and a recipe to build the above submodule:
>
> http://cgit.bec-systems.com/cgit.cgi/openembedded.custom/tree/recipes/autotools-demo/autotools-demo.bb
>
> ===================
> # recipe to build the ics-test-app in the working tree
>
> inherit srctree gitver autotools
>
> PV = "${GITVER}"
> S = "${OEDIR}/openembedded.custom/work/autotools-demo"
> ==================
>
>
> When I try 'bitbake autotools-demo', I get the following:
>
> cbrake at happy:/scratch/oe/oe-build/oe$ bitbake autotools-demo
> NOTE: Handling BitBake files: / (4602/8385) [54 %]ERROR: 'NoneType'
> object has no attribute 'startswith' while parsing
> /scratch/oe/oe-build/oe/openembedded.custom/recipes/autotools-demo/autotools-demo.bb
> NOTE: <type 'exceptions.AttributeError'>:'NoneType' object has no
> attribute 'startswith' while evaluating:
> ${@get_git_pv('/scratch/oe/oe-build/oe/openembedded.custom/work/autotools-demo',
> d)}
> NOTE: <type 'exceptions.AttributeError'>:'NoneType' object has no
> attribute 'startswith' while evaluating:
> ${GITVER}
> NOTE: <type 'exceptions.AttributeError'>:'NoneType' object has no
> attribute 'startswith' while evaluating:
> ${PN}-${EXTENDPE}${PV}-${PR}
> NOTE: <type 'exceptions.AttributeError'>:'NoneType' object has no
> attribute 'startswith' while evaluating:
> ${FILE_DIRNAME}/${PF}/${MACHINE}
> ERROR: 'NoneType' object has no attribute 'startswith' while parsing
> /scratch/oe/oe-build/oe/openembedded.custom/recipes/autotools-demo/autotools-demo.bb
> NOTE: Handling BitBake files: \ (8384/8385) [99 %]Command execution
> failed: Traceback (most recent call last):
>  File "/scratch/oe/oe-build/oe/bitbake/lib/bb/command.py", line 85,
> in runAsyncCommand
>    self.cooker.updateCache()
>  File "/scratch/oe/oe-build/oe/bitbake/lib/bb/cooker.py", line 802,
> in updateCache
>    if not self.parser.parse_next():
>  File "/scratch/oe/oe-build/oe/bitbake/lib/bb/cooker.py", line 995,
> in parse_next
>    raise ParsingErrorsFound
> ParsingErrorsFound
>
> Any suggestions as to what might be going wrong with gitver?
>
> Thanks,
> Cliff
>