Hi Tudor,

there are two types of problems in the jdt2famix-import.problems file.

My project uses Lombok to generate getter, setters and builders and so on (which is quite commmon for java projects). Because of that, some methods can't be found. Consider the following classes:

----------------------------------------
import lombok.Builder;

@Builder
public class Address {

private String postalCode;

}
----------------------------------------
import lombok.AccessLevel;
import lombok.NoArgsConstructor;

@NoArgsConstructor(access = AccessLevel.PRIVATE)
public class AddressBuilder extends Address.AddressBuilder {

public static AddressBuilder anAddressBuilder() {
return new AddressBuilder();
}

public Address.AddressBuilder withDefaultValues() {
return this.postalCode("23456");
}

}
-----------------------------------------

I get the following error:
unresolved method declaration - withDefaultValues - c:\data\moose\...\AddressBuilder.java - line 12

This error message and its line number (line number 12 is the method header "Address.AddressBuilder withDefaultValues ...") are misleading because at first I thought that the hole method is missing. But it only means that in this method an unresolved method is used. Furthermore I wonder if it were possible to support lombok?

The cause of the next problem is that we have multiple smaller applications (e.g. backoffice and frontend) which are separate applications but share a common db. Some of these services have classes with the same class and packagename (e.g. an annotation WebController). This lead to the following error message:

unresolved annotation type declaration - WebController - c:\data\moose\....\WebController.java - line 12

I had to start the debugger to find the issue. I found it because AnnotationTypeDeclaration.parent.problems = "Pb(323) The type WebController is already defined"

If the package names are different the problems disappears. Then I only get the the lombok errors. As a solution I will create unique base package names for my projecfs (which makes sense anyway).

Note that you could the following example to you README of jdt2famix to show how to download the dependencies for gradle projects:

task copyDependencies(type: Copy) {
from configurations.compile
from configurations.testCompile
into 'dependencies'
}

Best regards
Meinert

-----Original Message-----
From: Tudor Girba [mailto:tudor@tudorgirba.com]
Sent: Thursday, August 11, 2016 10:41 PM
To: Moose-related development
Subject: [Moose-dev] Re: Problems with mondorian

Hi,

Great. I am happy it works for you.

Still, could you tell me if you get any problems reported in jdt2famix-import.problems when you parse your system?

Cheers,
Doru

> On Aug 11, 2016, at 6:12 PM, Meinert Schwartau <m.schwartau@gmail.com> wrote:
>
> Thanks. After downloading the current Moose version and using the
> other method you suggested it works for me :-)
>
> Thank you for providing the Java parser.
>
> Best regards
> Meinert
>
> > Am 10.08.2016 um 16:36 schrieb Tudor Girba <tudor@tudorgirba.com>:
> >
> > Hi,
> >
> > I now fixed the providerTypes error (including a test). Please try with the latest Moose 6.0 image.
> >
> > Cheers,
> > Doru
> >
> >
> >> On Aug 10, 2016, at 3:49 PM, Tudor Girba <tudor@tudorgirba.com> wrote:
> >>
> >> Hi,
> >>
> >> Welcome back :)
> >>
> >>> On Aug 10, 2016, at 9:04 AM, Meinert Schwartau <m.schwartau@gmail.com> wrote:
> >>>
> >>> Hi,
> >>>
> >>> I want to display the dependencies between my classes. I wonder why the following code does not work, it displays the classes in a circle but not the edges between them. I’m using Moose 6 und Pharo 5 (downloaded yesterday) and evaluated the following code in the moose panel in the evaluator:
> >>>
> >>> |view|
> >>> view := RTMondrian new.
> >>> view nodes: ArrayedCollection withAllSubclasses.
> >>> view edges: (ArrayedCollection withAllSubclasses) from: [ :cls | cls yourself ] to: [ :cls | cls referencedClasses ].
> >>> view circleLayout.
> >>> view
> >>
> >> In your script, from:to: connects one source node with one target node returned by evaluating the corresponding blocks. However, your to: blocks return a collection, and the engine will try to find a node that has that collection as a model.
> >>
> >> What you want is to iterate over all items in the collection and connect the source node to each of the target nodes.
> >>
> >> To this end, you should use toAll:
> >>
> >> |view|
> >> view := RTMondrian new.
> >> view nodes: ArrayedCollection withAllSubclasses.
> >> view edges source: (ArrayedCollection withAllSubclasses) connectFrom: [ :cls | cls yourself ] toAll: [ :cls | cls referencedClasses ].
> >> view circleLayout.
> >> view
> >>
> >>>
> >>> Then I tried to display the dependencies between my own classes (parsed by jdt2famix) but got an exception. After clicking on All classes in the moose panel I entered the following code in the evaluator:
> >>> |view allClasses|
> >>> view := RTMondrian new.
> >>> allClasses := self allClasses.
> >>> view nodes: allClasses.
> >>> view edges: allClasses from: [ :cls | cls yourself ] to: [ :cls | cls providerTypes].
> >>> view circleLayout.
> >>> view
> >>>
> >>> If I execute the code above, I get an “MessageNotUnderstood: reveiver of “atScope:” is nil” exception. If I remove the “view edges: allClasses from: [ :cls | cls yourself ] to: [ :cls | cls providerTypes].” statement I don’t get an exception, the RTMondorian view opens, but no classes are displayed as dots in the view.
> >>
> >> Indeed, thanks for reporting.
> >>
> >> I also noticed a bug in MooseQuery during the computation of messages like providerTypes. The problem appears when the opposite part of a relationship is nil. For example, when you have an invocation and the target cannot be resolved, the invocation will point to nil, and Moose gets unhappy. This is a problem in Moose and we need to solve it.
> >>
> >> Cheers,
> >> Doru
> >>
> >>
> >>> Any suggestions?
> >>>
> >>> Best regards
> >>> Meinert
> >>>
> >>>
> >>> _______________________________________________
> >>> Moose-dev mailing list
> >>> Moose-dev@list.inf.unibe.ch
> >>> https://www.list.inf.unibe.ch/listinfo/moose-dev
> >>
> >> --
> >> www.tudorgirba.com
> >> www.feenk.com
> >>
> >> "If you can't say why something is relevant, it probably isn't."
> >
> > --
> > www.tudorgirba.com
> > www.feenk.com
> >
> > "From an abstract enough point of view, any two things are similar."
> >
> >
> >
> >
> >
>
> _______________________________________________
> Moose-dev mailing list
> Moose-dev@list.inf.unibe.ch
> https://www.list.inf.unibe.ch/listinfo/moose-dev

--
www.tudorgirba.com
www.feenk.com

"Not knowing how to do something is not an argument for how it cannot be done."