The Google Search plugin formats queries as: http://www.google.ca/search?q=%S&start=0&ie=utf-8&oe=utf-8&client=firefox-a&rls=org.mozilla:en-US:official But the Google Quicksearch bookmark uses: http://www.google.ca/search?q=%S and keyword.url uses: http://www.google.com/search?btnI=I%27m+Feeling+Lucky&ie=UTF-8&oe=UTF-8&q= We should probably unify these such that they are all using the search parameters: &client=firefox-a&rls=org.mozilla:en-US:official

Uhm... &ie=UTF-8&oe=UTF-8 were important arguments, because other user agents do not always use UTF-8. So google did not return proper answers without them, at most 3 years ago. It seems to me google might have fixed this issue at server side. I can't reproduce the problem now, there's no need to trim them though. Other params are dup of HTTP header's User Agent string, so I guess google would use (have used) them as of a debug purpose or the pure statistical analysis. I have no objection to add them all, but is there any chance to hear an opinion from google.com developers? Maybe they don't need such extra information. BTW, "&client=firefox-a" (official) or "&client=firefox" (unofficial) would triger the redirection (from google.com -> .ca, .fr, .co.uk etc.) See the difference if you live in other than US: http://www.google.com/search?q=foo&client=firefox (redirecting to google.**) http://www.google.com/search?q=foo&client=seamonkey (no redirection)

The Google quick search has been removed (bug 348165), and keyword.url was changed to use BBN (bug 323801). Is this FIXED, or are further tweaks needed to the BBN URL? Or is this only a 1.5.0.x branch bug?

At this point its 1.5.0.x and older, and I'm just going to mark it WONTFIX, since these codes are governed by contracts that don't cover these searches.