[kepler-dev] The 'poor man's Kepler Repository'

[Shawn Bowers]

Dan Higgins wrote:
 > Frankly, I don't see a lot of advantage in downloading kars until we
 > can  include data and java classes in the kars. [That doesn't mean we
 > shouldn't do it now in preparation for those capabilities.]

I agree -- we aren't going to get major benefits other than possibly
better organization/packaging (e.g., as you have done on the R in
Kepler webpage, which I think is nice -- ), and also maybe making it
easier for folks to incorporate multiple actors at once.  We won't see
major benefits though until we can decouple and support external java
actor classes.


-shawn

 >
 > Dan
 >
 > ---
 >
 > Shawn Bowers wrote:
 >
 >> Dan Higgins wrote:
 >>
 >>> Hi Shawn,
 >>>    I am not sure I see a lot of cases where one would want to move an
 >>> entire workflow to the actor library (and one cannot do that directly
 >>> right now even from a local workflow).
 >>
 >>
 >> Do you distinguish composite actors and workflows? If not, I think 
there
 >> are a number of reasons why someone would want to do this. Also, I
 >> believe Chad has an initial version of this working, i.e., the ability
 >> to add composites.
 >>
 >>> But if you wanted to (or just want to add some collection of actors
 >>> to the library), you can select the set of actors (including the
 >>> entire workflow), then select 'Tools/Create Composite Actor' and
 >>> combine the contents of the selection into a single composite. The
 >>> composite can then be put into the library.
 >>
 >>
 >> Doing this sounds more like lumping things together as opposed to 
having
 >> a functional unit (composite) ...
 >>
 >> -shawn
 >>
 >>>
 >>> Dan
 >>>
 >>> Shawn Bowers wrote:
 >>>
 >>>> Dan Higgins wrote:
 >>>> > Hi All,
 >>>> >     In yesterday's conferece call we spent some time discussing a
 >>>> Kepler
 >>>> > repository. Well, it turns out that you can share workflows 
using the
 >>>> > current system and a simple web html server. To see an example:
 >>>> >
 >>>> > 1)open Kepler
 >>>> >
 >>>> > 2) go to the menu 'File/Open URL..."
 >>>> >
 >>>> > 3) enter the url "http://nceas.ucsb.edu/~higgins/R_in_Kepler.html"
 >>>> >    Kepler will display the R_in_Kepler web page. (Actually it will
 >>>> > display most web pages; it is a simple web browser)
 >>>> >
 >>>> > 4) Click one of the links. The remote workflow will be opened.
 >>>> >
 >>>> > 5) You can then run the workflow (all these quick examples require
 >>>> R to
 >>>> > be installed). Or you can select an actor (one of the R actors) and
 >>>> > "Create KAR from actor..." to add it to the local machine (or save
 >>>> the
 >>>> > workflow)).
 >>>>
 >>>> Thanks Dan -- I think this is great.
 >>>>
 >>>> What would be cool is if we could get it to recognized kepler archive
 >>>> files (.kar files), and upon selection, kepler could prompt the user
 >>>> to either: (1) import the archive (i.e., load it's components into 
the
 >>>> library; (2) open the model (if it is a model); or (3) both.
 >>>>
 >>>> The current open url functionality allows us to open these external
 >>>> models, but we can't easily store them into the library ... (only the
 >>>> underlying actors).
 >>>>
 >>>> -shawn
 >>>>
 >>>> >
 >>>> > Dan
 >>>> >
 >>>>
 >>>
 >>>
 >>
 >
 >