[news.eclipse.amw] Re: Refactoring deprecated AM3 branch and AMW model h

[Hugo Bruneliere <hugo.bruneliere@xxxxxxxx>]

Hi Juan Pedro,

Juan Pedro Silva a écrit :
> Thank you very much, Hugo.
> I ended up with this piece of code (let me know if you consider it 
> appropriated):
>
>             try
>             {
>                 //AtlVM.getVM(vm).launch(asmUrl, libs, models, params, 
> superimps, options);
>                 //AtlLauncher.getDefault().launch(asmUrl, libs, models, 
> params, superimps, options);
>                 ASM asm = new ASMXMLReader().read(new 
> BufferedInputStream(asmUrl.openStream()));   
>                 superimps.add(0, asm); //New ILauncher.launch(..) method 
> takes a list of modules where the first module of the set is overriden 
> by the next ones
>                 ILauncher atlL = 
> CoreService.getLauncher(ATLLaunchConstants.REGULAR_VM_NAME);               
>                 atlL.launch(ILauncher.RUN_MODE, new 
> NullProgressMonitor(), options, superimps);
>             } catch (IOException e)
>             {
>                 ATLLogger.log(Level.SEVERE, e.getLocalizedMessage(), e);
>             } catch (ATLCoreException e) {
>                 ATLLogger.log(Level.SEVERE, e.getLocalizedMessage(), e);
>                 e.printStackTrace();
>             }
>
> I will do a partial test of my refactorization with an AMW 
> refactorization I'm also performing (it may take a while, though, as I 
> have other thing in hand).
> Perhaps (if you already have test suites for AM3) you may like me to 
> send it to you so you could start testing it already?. Just let me know.

Unfortunately, as we are not continuing the development of these 
plug-ins anymore, we don't have such a test suite. However, one simple 
way of validating your new implementation is to manually test each of 
the injectors, extractors, and other converters (concerning KM3, ATL, 
XML, etc) on a given set of files/models/metamodels. Thus, you will be 
able to ensure that the general behavior of these plug-ins has not been 
altered.

Similarly, for AMW, you can try to run some of the use cases available 
from the AMW website using both ATL 3.0 and your new implementation of 
the AMW model handler.

> How is everyone in EDM-Nantes?, send my regards.
> Merçi ;-),

Everything goes well at EMN. Thanks!

Cheers,

Hugo

>               Juan Pedro
>
>
> Hugo Bruneliere escribió:
> > Hi Juan,
> >
> > Juan Pedro Silva a écrit :
> > > Hi Hugo, thank you for your support.
> > > If I take the mentioned approach, the following instruction:
> > >
> > > AtlVM.getVM(vm).launch(asmUrl, libs, models, params, superimps, 
> > > options);
> > >
> > > Is no longer supported (AtlVM is no longer there in ATL 3.0), so I 
> > > need a suggestion in how to rewrite that instruction.
> > > These alternatives might work, but not sure which is best:
> > >
> > > RegularVMLauncher.launch(java.lang.String mode,
> > >                                
> > > org.eclipse.core.runtime.IProgressMonitor monitor,
> > >                                
> > > java.util.Map<java.lang.String,java.lang.Object> options,
> > >                                java.lang.Object... modules);
> > >
> > > or
> > >
> > > AtlLauncher.getDefault().launch(java.net.URL asmurl, java.util.Map 
> > > libraries, java.util.Map models, java.util.Map asmParams, 
> > > java.util.List superimpose, java.util.Map options)
> > >
> > > The first alternative is using the regular VM, but parameters are not 
> > > the same, and as documentation is scarce, I don't know how to convert 
> > > them.
> > > On the other side, the second alternative takes the same parameters, 
> > > but my feeling is that it will use the new VM (as it is the default 
> > > one), and not regular VM.
> > > There may be other (more correct) alternatives, but I'm not 
> > > acquainted with the new API, and those are the ones I found so far.
> > > Do you have any suggestions for this instruction?.
> >
> > In order to get the launcher of the Regular VM (but which conforms to 
> > the new ILauncher interface), you can use the "getLauncher()" method 
> > of the org.eclipse.m2m.atl.core.service.CoreService class:
> >
> >    ILauncher atlLauncher = 
> > CoreService.getLauncher(ATLLaunchConstants.REGULAR_VM_NAME);
> >    //of course, ATLLaunchConstants.EMF_VM_NAME also exists
> >    atlLauncher.launch(...);
> >     
> > Cheers,
> >
> > Hugo
> >
> > > I'll love to contribute once I finish refactoring.
> > > Thanks for your help.
> > > Regards,
> > >               Juan Pedro
> > >
> > > Hugo Bruneliere escribió:
> > > > Hi Juan,
> > > >
> > > > First, thank you very much for your contribution. I'm sure it will 
> > > > be used by many AM3 & ATL developers :-)
> > > >
> > > > Juan Pedro Silva a écrit :
> > > > > Hi everybody, I'm trying to rewrite some parts of AM3's deprecated 
> > > > > packages branch to use ATL 3.0.
> > > > > I have managed to modify most parts to use the new API, but I have 
> > > > > doubts on a particular piece of code.
> > > > > The code in hands is in class 
> > > > > org.eclipse.gmt.am3.tools.ant.ATLModelTransformationTask, and is 
> > > > > this one:
> > > > >
> > > > >         ...
> > > > >         if(vm == null)
> > > > >             vm = AtlVM.ATLVM_REGULAR;                  ...
> > > > >         if (!debug.booleanValue())
> > > > >             AtlVM.getVM(vm).launch(asmUrl, libs, models, params, 
> > > > > superimps, options);
> > > > >         ...
> > > > >
> > > > > I guess I can rewrite the first red line as
> > > > >            vm = ATLLaunchConstants.REGULAR_VM_NAME;
> > > > >
> > > > > but I don't believe this is the correct way to go, I think I should 
> > > > > change the structure somehow.
> > > > > Can anyone give me a pointer on this?.
> > > >
> > > > From my point of view, what you propose is good because it keeps 
> > > > exactly the same behavior as the previous version. I don't think it 
> > > > is necessary to modify the current structure of the plugin, at least 
> > > > in a first time.
> > > >
> > > > Of course, it would be great if you could contribute your upgrades 
> > > > by submitting a patch via Bugzilla!
> > > >
> > > > Cheers,
> > > >
> > > > Hugo
> > > >
> > > > > Thanks,
> > > > >             Juan Pedro


--
-------------------------------------------------
Hugo Bruneliere - R&D Engineer
AtlanMod research team (INRIA & EMN) - Room B206
École des Mines de Nantes
4, rue Alfred Kastler
44307 Nantes Cedex 3 - France
Office: +33 (0)2 51 85 82 21
EMail: hugo.bruneliere@xxxxxxxx
-------------------------------------------------