On 5 May 2010, at 11:30, Simon Denier wrote:
On 5 mai 2010, at 10:22, Tudor Girba wrote:
> MondrianPaintings holds Mondrian views which are typically 2-3
> methods each. Distribution Map, with all the Wizard and other
> computations is more than that, so I think it deserves a package on
> its own :)
I think like that for MondrianScripts, which are often class
extensions, but not for MondrianPaintings which often holds multiple
classes for the same application. Then CycleTable, LayerTable, and
PackageBlueprints have comparable size to DistributionMap. But I
like the fact that they all appear together in the browser.
Ok, for me MondrianScripts should go in MondrianPaintings
> Actually, I do not like categories :). I prefer to have packages
> for everything.
Well packages are for distribution, categories for internal
organization (so I don't mind them if this is built this way). It
makes sense to have multiple packages when:
- you end up with too many merges due to co-evolution of unrelated
things within a single package (I don't feel we don't have this
problem right now with Moose-MondrianPaintings)
In my opinion there are too many interests in this package already.
- you have independent applications mixed with Moose-dependent
applications: we could put DistributionMap, CycleTable, LayerTable
in MondrianPaintings (so the one from Mondrian, not Moose-...) since
they do not depend on Famix at all. But then we have to take care of
new dependencies for Mondrian (like Merlin)
The DistributionMap Wizard depends on Moose and on Fame not just on
Merlin, so they do not belong in Mondrian.
But, indeed I think that the DistributionMap visualization engine and
the other ones could belong to Mondrian as a layer on top.
In that case, the Distribution Map Wizard would be packaged in Moose-
What I am sure of is that I want to move MondrianScripts into
MondrianPaintings, because the name is better.
Does it sound right?
> On 4 May 2010, at 16:38, Simon Denier wrote:
>> On 4 mai 2010, at 10:36, Tudor Girba wrote:
>>> Hi Cyrille,
>>> I have a bit of feedback for Distribution Map:
>>> - The computation from Distribution Map Wizard is too expensive
>>> because you iterate over the meta elements and for each of them
>>> you go over all elements of Moose through the method
>>> It would be better if you would go over all cached groups and
>>> decide based on that rather than traversing the entire entities
>>> collection (which can be millions of objects).
>>> - It would be better if Distribution Map gets its own package
>>> because it starts to be large
>> What does large mean in this context? I mean, I don't understand
>> the benefit of creating a single package for DistributionMap while
>> all others are under MondrianPaintings (within their own categories)
>> Moose-dev mailing list
> "Every now and then stop and ask yourself if the war you're
> fighting is the right one."
> Moose-dev mailing list
Moose-dev mailing list
"Every thing should have the right to be different."