Is this is the spec yet? I'm assuming that we need to de-dupe the everything.me google result against this as well?

Yup its in the spec, and it is supposed to dedupe was decided sorry filing duplicates for myself https://mozilla.app.box.com/s/2tix674298wtc4e4hewh/1/1399872384/16143647367/1 page 30

Didnt do l10n and customisability as this is almost certain to change, implemented for user testing

So my preference here is not not tie in this specialized google logic into the collect() method. I'd like to try implementing this as a provider actually, albeit one that only returns a single result. If we do this then the de-duplicator should just work and the E.me result should be removed. It seems we may need a branch for user testing, so we may be able to land that there, but I'd like to try the provider route before landing in master. What do you think?

The problem is it behaviour different from the rest of the dedupe logic, the UX spec says it is shown when the e.me results contain the search result, and the search result is removed from the e.me results If this was a provider, it would either come before the e.me results and therefore show unconditionally, or it would come after and not show, I dont see any reasonable way to generalise it, its an exception in the UX spec.

I would say that we might be able to argue with UX about this - it seems that most of the E.me results come back with google.. Anyway - I still think that any result displayed should come from a provider. In this case the WebResults provider could actually spawn it's own "GoogleSearchProvider" class up, which points to the search-link container. This way the WebResults provider is in charge of parsing to determine if we have a google result or not - I think this is much cleaner than doing it in search.js. What do you think?

Refactored to be a provider managed by webresults, no changes to search.js

Comment on attachment 8408694 [details] https://github.com/mozilla-b2g/gaia/pull/18441 Thanks for putting up with me on this. I like the new solution because it keeps search.js small and should be easier to rip out in the future if we want. I just left one comment on github about a mystery googlelink.js file :)

Nah thanks, you were right this is much cleaner and will be easier to rip out when we hopefully do so :) https://github.com/mozilla-b2g/gaia/commit/47c95468061f61d96614177bd89edbe165b7d6a6