[gme-users] copy and paste references

[Akos Ledeczi]

The situation is even worse. Libraries are the "easy"
case, they are read-only. But what does a GUID mean
when you copy an object? Should it be preserved?
Clearly not, because is is not unique any more. Then
how would this help in copying a reference across
projects? The guids for the original referred object
and the new >>to be automatically resolved to<< one
is not same...

Even if you have some kind of id that works across
projects, what happens if you modify that model?
Is it still the same, so a copied reference should
still be resolved to refer to this object or not?



Aditya Agrawal wrote:

>Dear Peter,
>
>The first question is will the guid'd be preserved through the XML
>import/export? I would guess that they should.
>
>In GReAT we are using the library concept in a way where GUID's may not
>be able to help.
>
>In our cases the library mga file is not hand edited but generated. In
>this case from one version of the library to the other the guid's will
>not be preserved. In such a case we developed our own update tool that
>was aware of our paradigm and knew the attributed that could be used for
>update. For example, in the Uml paradigm the class names have to be
>unique and can be used to perform the update, similarly in MetaGME the
>(type, name) pair is unique and could be used. 
>
>I was wondering if we could develop a utility that can be configured to
>state the unique identifier of the paradigm and use it to perform the
>update.
>
>Please comment on these thoughts.
>
>Thanks,
>Adi
>
>------------------------------------------
>Aditya Agrawal 
>mailto:aditya.agrawal at vanderbilt.edu
>http://aditya.isis.vanderbilt.edu/
>Institute For Software Integrated Systems (ISIS) www.isis.vanderbilt.edu
>
>Vanderbilt University 
>2015 Terrace Place 
>Nashville, TN 37203
>615 322 3164
>
>
>-----Original Message-----
>From: Peter Volgyesi [mailto:peter.volgyesi at vanderbilt.edu] 
>Sent: Monday, February 23, 2004 4:16 AM
>To: gme-users
>Subject: Re: [gme-users] copy and paste references
>
>We are currently in the process of completely redesigning the library 
>feature. Your working scenario is one of the typical situation we'd like
>
>to address with the new implementation (eg.: we're introducing GUID-s 
>for all MGA objects, so merging could be smarter). However, we still 
>have bunch of problems, questions. Do not expect this new library code 
>within a month (or two...)
>--
>peter
>
>
>  
>
>>On Thursday, 19 February 2004, Akos Ledeczi wrote:
>>
>>    
>>
>>>Attila,
>>>
>>>this is >>not<< a general error, that is I can do what
>>>you described using my paradigm and it works properly,
>>>that is it copies references just fine without the
>>>restriction you mentioned.
>>>
>>>Could you please post a bug report on bugzilla
>>>specifying the paradigm you use and an action
>>>sequence that causes the crash?
>>>      
>>>
>>On the same topic, we have a small problem with the way models that
>>    
>>
>are
>  
>
>>built using the same single underlying imported library, are merged.
>>
>>The scenario we have is different people have to work on building
>>different
>>aspects of the same modeling paradigm, but everyone is using the same
>>library, say ESML. Assuming the common portions of the extended
>>    
>>
>library
>  
>
>>are
>>captured, the individual developers are free to extend their aspects
>>    
>>
>of
>  
>
>>the
>>paradigm. Now when the aspects need to be merged, it would be nice if
>>    
>>
>the
>  
>
>>copy/paste done as part of the merging, maintained the reference
>>information of elements of each aspect to elements defined in the
>>    
>>
>imported
>  
>
>>library, instead of
>>
>>1. Either invalidating the references because GME keeps track of
>>    
>>
>objects
>  
>
>>   via id which obviously fails when doing copy/paste. 
>>2. Having multiple copies of the same base imported library.
>>
>>I was wondering if it is possible to provide such support for atleast
>>MetaGME paradigm in GME, as the combination of kind and type should be
>>unique in MetaGME.
>>
>>We (Adi and me) discussed a bit on trying to solve it on the same
>>    
>>
>lines of
>  
>
>>the LibraryUpdate tool that he implemented. I am looking for arguments
>>for/against having such support in GME.
>>
>>Without this, we effectively have to serialize the process of creating
>>meta-models, which IMHO, is a waste of time.
>>
>>Comments/Suggestions very welcome.
>>
>>-kitty.
>>
>>_______________________________________________
>>gme-users mailing list
>>gme-users at list.isis.vanderbilt.edu
>>http://list.isis.vanderbilt.edu/mailman/listinfo/gme-users
>>    
>>
>
>_______________________________________________
>gme-users mailing list
>gme-users at list.isis.vanderbilt.edu
>http://list.isis.vanderbilt.edu/mailman/listinfo/gme-users
>_______________________________________________
>gme-users mailing list
>gme-users at list.isis.vanderbilt.edu
>http://list.isis.vanderbilt.edu/mailman/listinfo/gme-users
>  
>