OSX release somehow unstable (was Re: [CLAM] Re: CLAM digest, Vol 1 #235 - 2 msgs)

[Pau Arumi]

En/na Xavier Amatriain ha escrit:
> I have been testing the different apps on OSX and I also get different
> errors related to last minute packaging problems. On QtSMSTools I get a
> weird error (weird in the sense that it should also affect other OSs) of
> ports not being of the same type:
> 
>  /Volumes/CLAM_QtSMSTools/QtSMSTools.app/Contents/MacOS/QtSMSTools
> ##########################################################
> ################### ASSERTION FAILED #####################
> ##########################################################
> At file core/include/CLAM/InPortPublisher.hxx line 67
> InPortPublisher<Token>::PublishInPort coudn't connect to outPort because
> was not templatized by the same Token type as InPortPublisher
>  Unable to adquire symbols names for the backtrace
> terminate called after throwing an instance of
> 'CLAM::ErrAssertionFailed'
>   what():  InPortPublisher<Token>::PublishInPort coudn't connect to
> outPort because was not templatized by the same Token type as
> InPortPublisher

Xavier, I can't reproduce this problem.
Both the installed version (from the released dmg) and the
compiled version works fine for me.

You are testing the binary (dmg) versions of clam and apps,
right? If you are compiling the code, check that both libs and
apps are using release=yes.

The assert fails at load time or later ?

Do you have back-trace ? would be interesting to see.

The result of 'ldd QtSMSTools' could also help.

In case tomorrow we are still clue-less, I'll work on setting the 
unit-tests on mac (using libs), which tests the dynamic 
connection between ports.



> 
> On NetworkEditor, I do not experience the "hang" problem explained by
> Stephane but once the App is running I cannot add any Processing because
> of a problem with factory keys and Ladspa (why is it trying to
> instantiate a Ladspa Loader when I add a Generator?). I also noticed
> that the NetworkEditor is also trying to use the ALSA interface,
> shouldn't it be set to PortAudio on OSX?

No, AudioIn/Out uses CoreAudio on OSX. The binding is at
compile-time (using #ifdefs)

 > ##########################################################
 > ################### ASSERTION FAILED #####################
 > ##########################################################
 > At file core/include/CLAM/Factory.hxx line 126
 > GetCreator invoked with a non existent key : MonoAudioFileReader
 > Registered keys are:
 > LadspaLoader,
 >  Unable to adquire symbols names for the backtrace
 > terminate called after throwing an instance of
 > 'CLAM::ErrAssertionFailed'
 >   what():  GetCreator invoked with a non existent key :
 > MonoAudioFileReader
 > Registered keys are:
 > LadspaLoader,


> 
> /Volumes/CLAM_NetworkEditor/NetworkEditor.app/Contents/MacOS/NetworkEditor
>  Using ALSA input/output driver.
>   No externalizer will work.
> Scheduler set to Round Robin with priority 46...
> ##########################################################
> ################### ASSERTION FAILED #####################
> ##########################################################
> At file core/include/CLAM/Factory.hxx line 126
> GetCreator invoked with a non existent key : MonoAudioFileReader
> Registered keys are:
> LadspaLoader,
>  Unable to adquire symbols names for the backtrace
> terminate called after throwing an instance of
> 'CLAM::ErrAssertionFailed'
>   what():  GetCreator invoked with a non existent key :
> MonoAudioFileReader
> Registered keys are:
> LadspaLoader,

The problem have nothing to do with the LadspaLoader. It's just
that the processing factory is (almost) empty !
Again, I don't have any clue.
But, I'd try to clean the clam installation and
start over (both installing dmgs and from sources).

Pau