On Jun 25, 2007, at 7:38 PM, Ben Adida wrote: > Stefano Mazzocchi wrote: >> Agreed, this <iframe> idea is not that useful if google sends you >> to a >> crappy HTML table page instead of your pretty exhibit-enabled one. > > So Keith Alexander's been working on an RDFa importer for Exhibit > which should make it into v2, and it really seems like the ideal > solution to this problem: make an ugly HTML table marked up with > RDFa, let Google index that, and use Exhibit JS to transform that > ugliness into the beautiful lens'ed view.
> You can even make the HTML table not so ugly so your non-JS-enabled > friends can enjoy the data. All of this, and no data repetition. There are plenty of drivers to assure that data repetition is the norm, not the exception. Not the least of which is that it's exceptional for the HTML to be the master copy, but also because the tool chains are easier to juggle if you tollerate renderings. On the client/server boundary you get data repetition because the server wants to be nice to different consumers: humans, search engines, feed-readers, the disabled, json consumers, rdf consumers (aka other computers). That's an ordered list; stuff at the front will get done, stuff at the end less so. Note though that some adopters are required to serve the disabled. It seems odd to take the last one in the list and increase the complexity of the first one. My take this afternoon, is that exhibit is a tool just like all those php and perl scripts people have been writing for the last decade plus. It bridges from a data store to a human friendly view of the data. It is happenstance that it kind of reveals a computer friendly view of the data in the JSON (when it does). That it fails to serve even the second audience on the list is just a feature shortfall; a sign of it's youth. Serving humans well we would like to show some love for four additional consumers: search engine spiders, feed readers, those with disabilities, and rdf consumers. But the intensity of our desire to give some love to each of these drops off really really fast. (Which is why there are laws about disability access.) Personally I think we ought to have a service that helps exhibit authors to obtain an HTML rendering of their entire collection. We should then advise them to place that into their page; hiding it from various audiences as appropriate. It seems plausible that such a thing might be cobbled out of Fresno or Crowbar with a little help from exhibit. All the other solutions I've heard outlined are either too clever or are trying to pile on other goals. I'm fine with the HTML having RDFa embedded in it, but that seems like a nice to have feature that could be added in a second step. It seems very unlikely that it's a good thing to stop offering to serve the JSON to other computers. It seems risky to be dependent on a RDFa reader that isn't done yet. The complexity of the exhibit -> HTML rendering tool is sufficient to make it seem inappropriate to add additional features to it's manifest before we write even a first version. - ben _______________________________________________ General mailing list General@simile.mit.edu http://simile.mit.edu/mailman/listinfo/general
