On Thu, Oct 10, 2013 at 10:42 AM, Karen Coyle <[email protected]> wrote:
> Actually, I can find that one with this query: > http://openlibrary.org/search?q=complete+old+english+sheepdog Oops! Right you are. > but as you see, the search results display is missing the author - > probably because that is supposed to come from the work record in the > display, even though it appears to be also in the edition record. (And > this author doesn't show up in an author search.) And there's a lot more > that is weird here - like it shows only one subject "Old English > sheepdog" and when you click on it you get this unlinked record but not > the other one that has the same subject heading. And of course in the > edition display you can't see the subject heading for this one because > that is on the work record... honestly, I'd need a diagram to make any > sense of this, but the upshot is that we've got major problems with > links between editions and works, and search is definitely not precise. > Yes, works records are integral to making things hang together the way the software expects. The stale search indexes are a separate issue, but in addition to making things unfindable via the search box, also cause things like the works on the authors page and editions on the works page not be be populated correctly because they depend on search to find them. > I assume the dumps include the links, since that's where you are getting > this from. With a curl command against the API (which is the outer > limits of my ability to poke around) I see links from editions to works. > I don't see a way to retrieve a work record using the API... But I > assume there is also a link from works to editions? I'd love to > understand how this all is supposed to hang together, but don't want to > take up people's time. If anyone has drawn a diagram or whatever for > their own purposes, I'd like to see it. (Or if there is an explanation > elsewhere that I've missed.) > A simpler way to look at this stuff than using curl is to just add ".json" to the URLs in your browser, e.g. http://openlibrary.org/books/OL5204939M.json http://openlibrary.org/authors/OL2006058A.json http://openlibrary.org/works/OL3923242W.json http://openlibrary.org/subjects/old_english_sheepdog.json If you install JSONview (Firefox or Chrome), it'll pretty print the JSON so that it's easier to read. By looking at the returned data, you can tell what is linked directly (e.g. covers & authors for a work) versus what's generated dynamically from search (e.g. editions for a work). Tom
_______________________________________________ Ol-tech mailing list [email protected] http://mail.archive.org/cgi-bin/mailman/listinfo/ol-tech Archives: http://www.mail-archive.com/[email protected]/ To unsubscribe from this mailing list, send email to [email protected]
