RE: [mylar-dev] Interest calculation rules

["Mik Kersten" <beatmik@xxxxxxx>]

Hi Igor,

As soon as an element is selected or edited it becomes interesting.  Each
selection and edit after that increments the interest.  Each selection event
also decays the interest of the element, so that if you only select the
element once than a number of selections and edits later it will disappear
from a filtered view.  Predicted interest (e.g. errors, active search,
propagation of interest to parents) can also raise the interest of elements.
That's how a class becomes interesting if you select a member.  All of the
factors in this calculation can be adjusted, and are only roughly tuned in
0.4.  Also, some will be exposed with UI in the future, and things like the
decay rate can be made to adapt dynamically (as long as that doesn't end up
feeling too magical).

I think that the problem you're seeing is that newly created resources fail
to become interesting right away if you have Link with Editor turned off on
the package explorer.  Try turning it on and then creating a class or file,
it should pop into the model right away because a selection event is
generated.  This is a usability 'bug' and we need to do something about it.







> -----Original Message-----
> From: mylar-dev-bounces@xxxxxxxxxxx [mailto:mylar-dev-bounces@xxxxxxxxxxx]
> On Behalf Of Igor Fedorenko
> Sent: November 10, 2005 9:49 AM
> To: mylar-dev@xxxxxxxxxxx
> Subject: [mylar-dev] Interest calculation rules
> 
> Hi,
> 
> I apologise if this has been answered before but I am looking for
> information about different heuristics Malar uses to calculate degree of
> interest and reasons behind these heuristics.
> 
> Particular behaviour I am trying to understand at the moment is how
> Mylar treats new and modified resources. For example, newly created
> project is not considered interesting enough. Likewise, it takes awhile
> before mylar considers new/changes files interesting even though current
> set of interesting resources is empty.
> 
> --
> Regards,
> Igor Fedorenko
> 
> _______________________________________________
> mylar-dev mailing list
> mylar-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/mylar-dev