[Clam-devel] Re: [CLAM] multiple ports

[Pau Arumí]

El ds 16 de 05 de 2009 a les 13:47 +0200, en/na Natanael Olaiz va
escriure:
> El 05/16/2009 01:32 PM, Pau Arumí escribió:
> > As Dirk warned, after applying this patch, any old (without applying the
> > migration script) network will assert/crash NE with this message:
> >
> > ##########################################################
> > ################### ASSERTION FAILED #####################
> > ##########################################################
> > At file scons/libs/core/src/OutPortRegistry.cxx line 44
> > No out port named 'AudioOut'.
> > Try with: '0
> >
> > Given that Sources and Sinks are pervasive in clam networks (and not all
> > of them are in our svn) I think this is quite risky and I suggest to
> > take advantage of this change to add some robustness funcionality:
> >
> > Instad of asserting when a connection is not correct, the network could
> > open without that connection and give the user a warning about the
> > broken connections. This message alone would suffice for making the user
> > aware about how to fix the network.
> >   
> 
> That would be great!
> 
> > Additinonally, the message could suggest the user the command to run to
> > update the new networks - or make it runnable (via a button) in this
> > message window.
> >   
> To do it safely, that doesn't require a version number on every processing?

Easier: we could store the clam global svn revision number in both the
clamnetwork xmls and the networks-migration-commands.
When NE sees an error in the network (such as broken connections) it can
look both numbers and say "hey, your network is saved with a
clam-version that needs a migration, let's apply it and open it again"

P

> 
> 
> Best regards,
> Nael.
> > I'd wait for this functionality (NE opens with broken connections)
> > before applying the patch.
> >
> > P
> >
> >
> >
> >
> > El dv 15 de 05 de 2009 a les 19:44 +0200, en/na Pau Arumi va escriure:
> >   
> >> Thanks Dirk, I will review and apply the patch. 
> >> Let's move (well, continue) the discussion in clam-devel ml, though .
> >>
> >> P
> >>
> >> En/na Dirk Griffioen ha escrit:
> >>     
> >>> Hi All,
> >>>
> >>> Attached you will find a patch for giving AudioSource and AudioSink 
> >>> multiple ports.
> >>>
> >>> I tested it manually and ran the testsuite (there is a second patch 
> >>> for the testsuite).
> >>>
> >>> There is one caveat: it does not seem possible to retain the old 
> >>> naming scheme when loading from a .clamnetwork file where a port has 
> >>> the name AudioIn or AudioOut; instead, they're now numbered 
> >>> (source/sinkname.number):
> >>>
> >>> AudioSource.0
> >>> AudioSource.1
> >>> ... etc
> >>>
> >>> In the jack connections the dot becomes an underscore (AudioSource_1 
> >>> ... etc).
> >>>
> >>> So the clamnetworks need to be updated - attached you will find an 
> >>> commandfile for the clamrefactor.py script.
> >>>
> >>> The reason for the namechange is that ports are constructed with a 
> >>> name which should match with the name in the xml file, but at 
> >>> construcion time this name from the xml file is not known. This went 
> >>> ok because the code always had 'AudioIn/Out' and this always matched 
> >>> with the xml file, as there was but one port and one portname for a 
> >>> source or sink.
> >>>
> >>> An alternative might be to create 'unnamed' ports (add a constructor, 
> >>> the default arguments are in the way), add a 'setname' function to the 
> >>> port and then call this when the xml is loaded. But it seems to me 
> >>> this would add a lot of complexity.
> >>>
> >>> When there is only 1 port in the source/sink, the jackport is named 
> >>> after the source/sink so when you have meaningfull names in your 
> >>> clamnetworks, these show in the jack connection viewer.
> >>>
> >>> I tested it and it looks good, but as I am new to CLAM I might missed 
> >>> (obvious) things; so please review.
> >>>
> >>> Best, Dirk
> >>>
> >>>
> >>>
> >>>
> >>>
> >>>
> >>>
> >>>
> >>>
> >>>
> >>> ------------------------------------------------------------------------
> >>>
> >>> _______________________________________________
> >>> CLAM mailing list
> >>> CLAM at iua.upf.es
> >>> http://clam.iua.upf.edu
> >>>       
> >> _______________________________________________
> >> CLAM mailing list
> >> CLAM at iua.upf.es
> >> http://clam.iua.upf.edu
> >>     
> >
> >
> > _______________________________________________
> > Clam-devel mailing list
> > Clam-devel at llistes.projectes.lafarga.org
> > https://llistes.projectes.lafarga.org/cgi-bin/mailman/listinfo/clam-devel
> >
> >   
> 
> 
> _______________________________________________
> Clam-devel mailing list
> Clam-devel at llistes.projectes.lafarga.org
> https://llistes.projectes.lafarga.org/cgi-bin/mailman/listinfo/clam-devel