[OE-core] Sad story about Shared State

[Richard Purdie]

On Fri, 2011-05-27 at 07:39 +0200, Martin Jansa wrote:
> On Fri, May 27, 2011 at 12:13:35AM +0100, Richard Purdie wrote:
> I meant something like switch back to something like pstage was or
> disable both (IIRC I had strange error on many places without sstate
> inherited by distro).
>  
> > Which BB_SIGNATURE_HANDLER are you using? It sounds like you're using
> > basichash but want the behaviour generated by basic (the current
> > default). Nobody is forcing you to use basichash.
> 
> AFAIK I'm using default basic..
> OE @ ~/shr-core/meta-smartphone $ git grep BB_SIGNATURE_HANDLER
> OE @ ~/shr-core/meta-openembedded $ git grep BB_SIGNATURE_HANDLER
> OE @ ~/shr-core/openembedded-core $ git grep BB_SIGNATURE_HANDLER
> meta/conf/bitbake.conf:BB_SIGNATURE_HANDLER ?= "basic"
> 
> OE @ ~/shr-core $ bitbake -e shr-lite-image | tee -a log; grep BB_SIGNATURE_HANDLER log
> # BB_SIGNATURE_HANDLER=basic
> BB_SIGNATURE_HANDLER="basic"
> 
> but even when I set it to not existing basic2
> it doesn't show error from
> lib/bb/siggen.py
>         logger.error("Invalid signature generator '%s', using default 'noop'\n"
>                      "Available generators: %s",
>                      ', '.join(obj.name for obj in siggens))
> but fails like this:
> NOTE: Preparing runqueue
> ERROR: Running idle function
> Traceback (most recent call last):
>   File "/usr/lib64/python2.7/site-packages/bb/server/process.py", line 151, in idle_commands
>     retval = function(self, data, False)
>   File "/usr/lib64/python2.7/site-packages/bb/cooker.py", line 792, in buildTargetsIdle
>     retval = rq.execute_runqueue()
>   File "/usr/lib64/python2.7/site-packages/bb/runqueue.py", line 940, in execute_runqueue
>     self.rqexe = RunQueueExecuteScenequeue(self)
>   File "/usr/lib64/python2.7/site-packages/bb/runqueue.py", line 1434, in __init__
>     valid = bb.utils.better_eval(call, locs)
>   File "/usr/lib64/python2.7/site-packages/bb/utils.py", line 390, in better_eval
>     return eval(source, _context, locals)
> TypeError: expected a character buffer object
> 
> I'll add some debugging code to bitbake to see what's really going on here..
> 
> > As you mention, there are benefits and drawbacks either way.
> 
> Yes I see the benefits and if we find out that basichash is somehow used
> instead of configured basic I'll be really happy to continue using it.

I dug into this with Martin on irc and it was found that rm_work.bbclass
was exposing a bitbake bug with setscene handling. This was fixed with:

http://git.openembedded.net/cgit.cgi/bitbake/commit/?id=0a4d857aabc86b973170ba9ce32b6b449a4e2165

and I think Martin is finding things work better now! :)

Cheers,

Richard