Hi Ted, Good luck; I like your ship listing.
I think perhaps you might be mixing up the members of your data set -- in your case, all 289 ships -- with facets. Facets are merely "slicing-and-dicing" dimensions or filters for choosing *subsets* from within your total data set. If no facet is chosen, it will default to "all", or the full listing of your 289 ships. So, in your case, facets might be decade built, shipyard, propulsion method, hull type, length of service, etc., etc. You can best judge those dimensions. Choosing these dimensions is really what is fun about Exhibit and gives it its browsing and display value. It is generally good practice to characterize each of your items in the data set (ships) via all facets, but that is also not strictly required (you may not have the data for that facet, may be its ambiguous, etc.). As for data set sizes, I know that David generally states around 500, but frankly I think mine performs well at that level and I am not concerned going twice that big. I think the more important performance aspect is whether you have an image associated with each data item or not. Mike Ted Bell wrote: > Thanks David, that does answer my question. > > Thanks for ex:showAll="false" (I had been wondering about something like > that); css is underway; and lens templates are under the microscope. > > I must admit to be in a quandary about the shipnames - there's not much > getting around the existence of 289 ships; it's kind of like saying that > there are 289 nobelists or Presidents, but you can't look them up by name > only. Mike Bergman's Sweet Tools has 500 data items but the name of a tool > doesn't need to be (or at least, isn't) a facet. I feel like there is > something that I am missing - either that or in my case I just happen to > need a facet with 289 choices. Are there any other examples in the wild that > would give me an idea of how else I might manage this? > > Ted > > -----Original Message----- > David Huynh said: > > Exhibit is good for about 50 - 500 items. 50 items are more than a person > can visually process and mentally keep track of, and so faceted browsing > starts to pay off. But these numbers are not based on any serious profiling, > and much depends on each user's browser and machine, too. There are ways to > improve performance: > - avoid facets that have low counts and thus many choices, e.g., the > Shipname facet has only "1" for each choice and so in total it has 289 > choices > - set ex:showAll="false" on the tile view so that only the first 10 items > are shown at the beginning > - make the lens templates simpler, if possible; use CSS classes instead of > inline styles And I'm speeding up various parts of Exhibit in version 2.0. > The bottleneck is in the dynamic UI generation and then in the faceted > browsing calculations. > .... > > Have I answered your questions? > > David > > > Ted Bell wrote: >> Greetings >> >> Exhibit's facetted browsing is fantastic and I've managed to create a >> small example: >>>> http://bbpress.navyhistory.org.au/nhsaships09.htm >> >> The size of the JSON file is 142KB (289 ships); an earlier version >> with just 14 ships was 7KB. >> I'd like to add data for 2 additional facets - "era" & "major battles" >> - I guess that would bump the file size up to 160-odd KB. >> I took the graphics out of the main tile page in order to speed up >> loading. >> >> Question: >>>> Is there a sweet spot for data volumes/properties for Exhibit? >>>> At what stage (of data volumes) should users stop using Exhibit and >> start using Longbow? >>>> Which brings me to... why develop 2 products (Longbow and Exhibit) >> that "seem" do the same thing (albeit with different data volumes)? >> >> Ted Bell >> _______________________________________________ General mailing list [email protected] http://simile.mit.edu/mailman/listinfo/general
