- From: Aryeh Gregor <Simetrical+w3c@gmail.com>
- Date: Thu, 14 Oct 2010 14:09:01 -0400
On Thu, Oct 14, 2010 at 12:12 AM, Peter Kasting <pkasting at google.com> wrote: > The API is completely orthogonal to OpenSearch AFAICT. ?OSDDs tell a UA > about a search engine and how to search with it. ?The proposed DOM additions > provide a search engine with information about what the user is doing in the > browser. My understanding of how this is meant to work is 1) When user starts typing in the browser search box, if the search engine supports it, the browser loads the search engine home page (or maybe some other designated page?) and displays it instead of the current page, as long as the user is typing a search. 2) The page runs like any other page, and uses the proposed searchBox API to figure out what the user has currently typed. This API is inaccessible unless the page was accessed via (1) (i.e., navigator.searchBox won't do anything if I just navigate to google.com from a bookmark). 3) The page updates to show search preview results, or whatever else it wants to do. If this is meant to be vendor-neutral, there needs to be some way for arbitrary search engines to advertise support for this feature to supporting browsers. In the experimental Chrome implementation, does it just not work if the search engine isn't Google? If Bing wanted to make use of this too, how would they do it? If advertising support is necessary, then OpenSearch seems like the way to do it. But I might be misunderstanding something here, since the proposal only gave an API, and didn't explain conformance requirements for UAs. So maybe my inferences from the demo video are wrong.
Received on Thursday, 14 October 2010 11:09:01 UTC