This started today and is reproducing currently on OS X machines only, with Nightly - fr and it locales. I'll check to see if it reproduces on my linux as well and if not prepare a skip patch for os x, just in case it escalates quickly.

This is now on all platforms, found bug 738818 is the cause as it's removing keyword.URL pref that we use in setup. Skipping patch attached until we figure out how we want to handle this test now.

Skipped: http://hg.mozilla.org/qa/mozmill-tests/rev/b213dacee111 (default)

Looks like a regression from a feature change. We should adapt our tests accordingly. If possible we should flip a sprint entry.

The feature our test checked, no longer exists. The preference was supposed to let us open a link (searching for the word) by just appending the word we typed in the location bar to the link given in the pref. Since that's gone, I found another option here [1], but is not working - I still get the Google search (default) when typing Mozilla Firefox in the locationbar. My proposal is to install/pick a search engine and make sure when we type in location bar opens that search. I used testAddMozSearchProvider.js from testSearch folder to do so and it works as expected. Or we can remove the test if the main purpose was to test the preference. [1] http://kb.mozillazine.org/Using_keyword_searches#Creating_bookmarks_with_keywords

Why do we have to install another search engine here? Lets use the one which is currently selected. Or what is the current test performing? Does it select another engine?

The default one is Google and the current test opens the google page, the location bar url is something like: https://www.google.ro/search?client=ubuntu&channel=fs&q=mozilla&ie=utf-8&oe=utf-8&redir_esc=&ei=upNmUY70F6SC4gTu_IDQCQ and test fails at getting the element with 'term' id (that was from the local page): http://www.mozqa.com/data/firefox/search/searchresults.html needed for assert. If we use this, we would need then to assert that locationbar url (google one) contains the string we gave. Our search engine from testAddMozSearchProvider.js is using searchresults.html page, but we would need to install and select to use it.

I haven't found any mochitests on mxr with this specific check, the ones mentioned in bug 738818 comment 17 were only deleting the code regarding the preference, since it was removed, so it won't fail. Added the check in testAddMozSearchProvider.js and removed the awesomebar test.

Forgot the reports: OS X: http://mozmill-crowd.blargon7.com/#/functional/report/cb9e33213cf3d2f37d7b0f8dcca1cd30 Windows: http://mozmill-crowd.blargon7.com/#/functional/report/cb9e33213cf3d2f37d7b0f8dcca2218a Linux: http://mozmill-crowd.blargon7.com/#/functional/report/cb9e33213cf3d2f37d7b0f8dcca4054c

(In reply to Andreea Matei [:AndreeaMatei] from comment #7) > I haven't found any mochitests on mxr with this specific check, the ones > mentioned in bug 738818 comment 17 were only deleting the code regarding the > preference, since it was removed, so it won't fail. Gavin, I'm asking just for sanity. Do we have any kind of mochitest in the tree which exercises typing in the location bar and letting it search with the default search engine? If yes, could you please reference the tests? Thanks.

(In reply to Henrik Skupin (:whimboo) from comment #9) > Gavin, I'm asking just for sanity. Do we have any kind of mochitest in the > tree which exercises typing in the location bar and letting it search with > the default search engine? Yes, http://mxr.mozilla.org/mozilla-central/source/browser/base/content/test/browser_keywordSearch.js does that.

Oh, good news then, thanks Gavin. Attaching patch to remove our test.

(In reply to :Gavin Sharp (use gavin@gavinsharp.com for email) from comment #10) > Yes, > http://mxr.mozilla.org/mozilla-central/source/browser/base/content/test/ > browser_keywordSearch.js does that. This test is still using keyword.URL to select the engine to use. I thought that has been removed? Why do we still exercise this path?

Comment on attachment 736691 [details] [diff] [review] patch v2 Review of attachment 736691 [details] [diff] [review]: ----------------------------------------------------------------- We should wait until it's clear what the state of the mochitest really is.

(In reply to Henrik Skupin (:whimboo) from comment #12) > This test is still using keyword.URL to select the engine to use. No, it's not. There is some now-unused code that wasn't cleaned up related to keyword.URL, but it's not affecting what the test actually checks. I removed it on inbound: https://hg.mozilla.org/integration/mozilla-inbound/rev/62db06ab5e42

Comment on attachment 736691 [details] [diff] [review] patch v2 Requesting review, giving Gavin's answer.

Comment on attachment 736691 [details] [diff] [review] patch v2 Review of attachment 736691 [details] [diff] [review]: ----------------------------------------------------------------- Looks good. Get rid of the test then. Thanks.

Landed: http://hg.mozilla.org/qa/mozmill-tests/rev/1f8005a4db12 (default)