[Clam-devel] Re: "Connect To" refactoring and bug

Natanael Olaiz nolaiz at gmail.com
Fri Jun 20 17:39:58 PDT 2008


After unconnecting the SMSAnalysisCore in port and try to reconnect it 
later, Network::ConnectPorts is returning with false because 
Network.cxx:443:

    if( inport.GetVisuallyConnectedOutPort())

is true.

With a new SMSAnalysisCore it works good.

Any idea? Some bug in disconnection management or the specific 
processing methods (in my tests seems to only have problems with 
SMSAnalysisCore)?



Regards,
Natanael.

El 06/20/2008 01:29 PM, Natanael Olaiz escribió:
> El 06/20/2008 12:28 PM, Pau Arumí escribió:
>> On dv, 2008-06-20 at 16:57 +0200, David García Garzón wrote:
>>  
>>> On Divendres 20 Juny 2008, Pau Arumí Albó wrote:
>>>    
>>>> El dv 20 de 06 de 2008 a les 06:23 -0300, en/na Natanael Olaiz va
>>>>
>>>> escriure:
>>>>      
>>>>> It seems to be there since first implementation.
>>>>> I tested r11328, the same example, the same results trying to add an
>>>>> AudioSource from SMSAnalysisCore_0, and others loaded processings 
>>>>> boxes
>>>>> (if you try with a new object, it works).
>>>>>         
>>>> seems a problem in the lowlevel IsConnectable()
>>>> note that you can neither connect the input to existing audio-source
>>>> (nor removing and redoing a connection)
>>>>
>>>> beyond that, clueless...
>>>> P
>>>>       
>>> I just did a heavy refactoring of the connect-to code. The main 
>>> refactor has been changing the method pointer by helper functions 
>>> that receive the region as parameter and call the proper method. The 
>>> result is slower but the code seems a lot clearer without 120 
>>> character declarations and without member indirections.
>>>     
> What a nice refactoring!
>
>>> I fixed a bug (which i don't know that was the bug you are refering 
>>> to or another one i introduced at some point) regarding the indexes 
>>> that were sent to the slot. They were swapped so if the index were 
>>> the same (ie. 0) it worked perfectly. So try the same steps to 
>>> reproduce it just in case it is still there. I have no problems now 
>>> with SMSAnalysisCore.
>>>     
> Still doing the same :(
> After open any example with a SMSAnalysisCore, if you unconnect the 
> audio port info and try to link to a new AudioSource, it creates the 
> audiosource but not the connection (is not a refresh wires problem).
>
> Further info: if after that you try to open the same/another example 
> using SMSAnalysisCore, NE quit with:
>
>    pure virtual method called
>    terminate called without an active exception
>
> Moreover: just unconnecting the SMSAnalysisCore from the original 
> AudioSource and trying to reload the example do the same. :(
>
>>> I also found a bug that could have confused you. The connection slot 
>>> didn't call to update after setting the connection, so that no wire 
>>> is drawn until the canvas is redrawn by other means as moving the 
>>> mouse into a processing. So it could happen that after doing a 
>>> connection the connection didnt show anything for a while.
>>>
>>> Further refactoring could be needed. Just that it was harder to 
>>> explain this one than actually coding it.
>>>     
>>
>> Sounds just great!
>> P
>>
>>
>> _______________________________________________
>> Clam-devel mailing list
>> Clam-devel at llistes.projectes.lafarga.org
>> https://llistes.projectes.lafarga.org/cgi-bin/mailman/listinfo/clam-devel 
>>
>>
>>   
>
>





More information about the clam-devel mailing list