Hi,On Fri, Oct 21, 2011 at 12:07 PM, Usman Bhatti <usman.bhatti@gmail.com> wrote:
On Fri, Oct 21, 2011 at 11:43 AM, Tudor Girba <tudor@tudorgirba.com> wrote:Hi,On Fri, Oct 21, 2011 at 11:16 AM, Usman Bhatti <usman.bhatti@gmail.com> wrote:
On Thu, Oct 20, 2011 at 11:19 AM, Tudor Girba <tudor@tudorgirba.com> wrote:Hi,On Thu, Oct 20, 2011 at 10:53 AM, Usman Bhatti <usman.bhatti@gmail.com> wrote:
Hello Doru,While porting the VW version of the lattice construction algos, the most adequate representation for VW data structures was provided by Mondrian so I used it.Mistakes happen. But, it is important to learn from them :)Now, I'll place them in Moose-Algos-Graph...I am not sure what you mean. I did not mean that you should move the Lattice classes in Moose-Algos-Graph. The Graphs should not depend on FormalConceptAnalysis. I was just saying that the Lattice is merely a support for creating a graph out of the FCA algorithm. In the VW version, it was also a visual class, but it does not have to be. It should only contain the logic of building the graph. Maybe for this purpose you can use the Graph concepts. The display can be later build using a Mondrian script.Ok, so I can break lattice drawing into two parts: construction + display.Yes.The display part can be moved to mondrian as an extension of FCA. So, that will break the dependency of FCA with Mondrian.No. Like I mentioned before, the display should be moved to Moose-MondrianPaintings. Mondrian is generic, and FCA is generic. They should not know about each other.Ok. That's a good place for lattice display. Is it normal that I have Moose-MondrianPaintings-* but there is no Moose-MondrianPaintings package in my image?At the moment, it is acceptable, but in the near future it won't be :).The lattice construction algo can be moved into Moose-Algos-Graph: Moose-Algos-Graph classes can be used to construct lattices instead of MONode. Now, lattice construction is not a simple graph construction and we need knowledge of concepts to construct a lattice. So, lattice construction would always depend on the concepts and then there is a dependency with the package FormalConceptAnalysis. I do not see a way that the lattice construction can work independent of FormalConceptAnalysis. Can you suggest a possible workaround for this?I did not say that it the construction should be independent. I simply said that if needed you might benefit from the graph support. In any case, the lattice will depend on FCA. So, either you place the Lattice class in the FCA package, or we create another package that will depend on FCA and possibly the graph one. The lattice does certainly not belong in the Graph package.Ok I want to use MalGraphAlgorithm and MalGraphNode as superclasses for Lattice and LatticeNode respectively, from Moose-Algos-Graph package.
Exactly.Now, I think creating a third package would be a better option because that would avoid the dependency:FormalConceptAnalysis <-> Moose-Algos-Graph.Yes.Should I create a package Moose-Algos-Graph-Lattice or Moose-Algos-Lattice?Please name it Moose-Algos-Lattice.If you name it Moose-Algos-Graph-Lattice, then Monticello will think it is a subcategory of Moose-Algos-Graph, due to implicit naming conventions. Actually, it is this implicit mechanism that makes it possible to have multiple Moose-MondrianPaintings-* categories and only one package. However, in Moose we want to keep as much as possible the 1-to-1 mapping between categories and packages.Cheers,DoruDoruOk for the tests. I'll start creating them now.Please put them in the newly created package (Moose-Tests-Algos-FormalConceptAnalysis).DoruI'll commit another version in the evening.thanxUsmanOn Wed, Oct 19, 2011 at 11:08 PM, Tudor Girba <tudor@tudorgirba.com> wrote:
I looked a bit more. The code from Lattice and LatticeNode depends on the logic of Mondrian. This should best be replaced with code that exists in Moose-Algos-Graph.
In the meantime I also created the Moose-Tests-Algos-FormalConceptAnalysis package and added it to the configuration.
Cheers,
Doru
--
On 19 Oct 2011, at 22:36, Tudor Girba wrote:
> Hi Usman,
>
> Please read this email. You just committed another version that broke the build :)
>
> I fixed it again.
>
> Cheers,
> Doru
>
>
>
> Begin forwarded message:
>
>> From: Tudor Girba <tudor@tudorgirba.com>
>> Subject: fca should not depend on mondrian
>> Date: 19 October 2011 21:23:09 CEST
>> To: Moose-dev Moose Dev <moose-dev@iam.unibe.ch>
>>
>> Hi Usman,
>>
>> I saw that you added the port code to the MooseAlgos repo. Thanks.
>>
>> Now, during your changes, you inherited the Lattice and the Node from MONode.
>>
>> This caused Moose to be unloadable and it is wrong from two points of view:
>> 1. MONode is not meant to be inherited at all. You are supposed to use Mondrian as a transformation from your domain into the internal Mondrian model.
>> 2. MooseAlgos is a basic set of algorithms with no dependencies. Mondrian is higher level. Thus, MooseAlgos cannot depend on Mondrian.
>>
>> For these reasons, I now removed the inheritance from MONode.
>>
>> Furthermore, we also need tests for FCA.
>>
>> Cheers,
>> Doru
>>
>>
>> --
>> www.tudorgirba.com
>>
>> "No matter how many recipes we know, we still value a chef."
>>
>>
>>
>>
>>
>>
>
> --
> www.tudorgirba.com
>
> "One cannot do more than one can do."
>
>
>
www.tudorgirba.com
"Some battles are better lost than fought."
_______________________________________________
Moose-dev mailing list
Moose-dev@iam.unibe.ch
https://www.iam.unibe.ch/mailman/listinfo/moose-dev
_______________________________________________
Moose-dev mailing list
Moose-dev@iam.unibe.ch
https://www.iam.unibe.ch/mailman/listinfo/moose-dev
--
--"Every thing has its own flow"
_______________________________________________
Moose-dev mailing list
Moose-dev@iam.unibe.ch
https://www.iam.unibe.ch/mailman/listinfo/moose-dev
_______________________________________________
Moose-dev mailing list
Moose-dev@iam.unibe.ch
https://www.iam.unibe.ch/mailman/listinfo/moose-dev
_______________________________________________
Moose-dev mailing list
Moose-dev@iam.unibe.ch
https://www.iam.unibe.ch/mailman/listinfo/moose-dev
_______________________________________________
Moose-dev mailing list
Moose-dev@iam.unibe.ch
https://www.iam.unibe.ch/mailman/listinfo/moose-dev
_______________________________________________
Moose-dev mailing list
Moose-dev@iam.unibe.ch
https://www.iam.unibe.ch/mailman/listinfo/moose-dev