There are security concerns in allowing to inline autocomplete pasted partial urls, since a user may paste a partial url and click enter without noticing the url he is going to is actually different from what he expected.

we may need a custom cmd_paste controller to detect this, similar to the _copyCutController one... I don't have better ideas atm, so any alternative idea is welcome.

I don't have better ideas atm, this works as far as the autocomplete fills up synchronously, that is not always true (if we fail matching on a domain it's likely a subpage and autofill for that happens asynchronously). So it's not yet the solution, it's a possible hook point though, that I would like some feedback or ideas on. Maybe we just care about not autocompleting domain matches and not subpages? in such a case it would be enough. Otherwise we could maybe store the last pasted value and compare with it in the first textValue setter call. It's still sorta hackish though.

How about differentiating between input events that add a single character and those that add multiple characters?

Comment on attachment 588430 [details] [diff] [review] hack >diff --git a/browser/base/content/urlbarBindings.xml b/browser/base/content/urlbarBindings.xml >+ <field name="_pasteController"><![CDATA[ >+ isCommandEnabled: function(aCommand) { >+ return this.urlbar.inputField.editor.canPaste(Ci.nsIClipboard.kGlobalClipboard); The nsPasteCommand::isCommandEnabled implementation seems to also check editor.isSelectionEditable, I wonder whether that's necessary? This doesn't really seem like a hack to me. It should be reliable, right? It would be nice to fix this problem for all autocomplete widgets, though (wait longer before starting the autocomplete searches triggered by pastes).

would something like this be fine? It delays autocomplete on paste by 1 second. I should still make a test.

I tried making a chrome mochitest witha type=autocomplete textbox, but the controller complains inputField.editor is null, I wonder if there's a privileges problem, the scope looks correct.

now it works, moving on.

There are still some things to define: - if the idl changes may be fine - if the delay should be configurable and how (pref/attribute)

Comment on attachment 589169 [details] [diff] [review] patch v1.0 new patch coming, according to irc discussion with Gavin.

Comment on attachment 589505 [details] [diff] [review] patch v2.0 >diff --git a/toolkit/content/widgets/autocomplete.xml b/toolkit/content/widgets/autocomplete.xml >+ <field name="_pasteController"><![CDATA[ >+ _clipboard: Components.interfaces.nsIClipboard.kGlobalClipboard, "_kGlobalClipboard"? just "_clipboard" is misleading.

addressed comment https://hg.mozilla.org/integration/mozilla-inbound/rev/19a5e75b8ed8

This seems to have regressed; filed bug 1022399.