Similarly to the browser_startup.js test landed in bug 1358798, we should test what scripts get loaded when starting a content process.

Fixed the eslint failure by enabling the frame-script environment inside the framescript, and disabled on !e10s where this doesn't make sense (as we can't create a new process so the test sees modules loaded by previous tests...).

Comment on attachment 8883386 [details] [diff] [review] Patch v2 Review of attachment 8883386 [details] [diff] [review]: ----------------------------------------------------------------- Thanks! ::: browser/base/content/test/performance/browser_startup_content.js @@ +40,5 @@ > + let tab = await BrowserTestUtils.openNewForegroundTab({gBrowser, > + forceNewProcess: true}); > + > + let mm = gBrowser.selectedBrowser.messageManager; > + let promise = new Promise(resolve => { Alternatively, waitForMessage from BrowserTestUtils could be modified to resolve with the message data. I wouldn't block on that though. http://searchfox.org/mozilla-central/rev/a3a739de04ee6134c11546568a33dbb6a6a29907/testing/mochitest/BrowserTestUtils/BrowserTestUtils.jsm#908

Pushed by florian@queze.net: https://hg.mozilla.org/integration/mozilla-inbound/rev/2d7a625e8c99 add a test preventing us from loading scripts unintentionally during content process startup, r=mconley.