Justin Dolske [:Dolske] Reporter
	Description • 16 years ago

Search history from the search box is simply stored in Satchel with a field name of "searchbar-history". This has two unfortunate side effects:

1) You can't easily clear form history and search bar history separately, since they're not quite independent.

2) If content uses <input name="searchbar-history">, history for that field will be mixed in with search box history.

The reverse of #2 seems a bit disturbing -- search box history will be available in content. I'm not sure this is a huge problem, because we already share form history across sites (it's only indexed by field name). Perhaps having this span content/chrome is too much, though.

OTOH, at some point in the future I'd like to have smarter form history matching -- aka awesomebar -- and it seems like having searchbox history available for that would be useful.

	Johnathan Nightingale [:johnath]
	Comment 1 • 16 years ago

yeah, combined with one of Jesse's "navigate through this maze while I steal your auto-fills" PoCs, this is potentially a significant privacy breach.

	:Gavin Sharp [email: gavin@gavinsharp.com]
	Comment 2 • 16 years ago

Bug 270697 made stealing this data a little bit harder, but I was wrong in assuming that it was now impossible, given bug 381681.

	Justin Dolske [:Dolske] Reporter
	Comment 3 • 13 years ago

Probably the easiest short-term (!) fix here is to just blacklist web site form autofill when the field name == out special chrome field name.

In the long term I'd still like to be able to have form inputs suggest completions regardless of origin, but that's a security issue we can take up if/when that feature ever tries to land.

	Matthew N. [:MattN] Assignee
	Comment 5 • 13 years ago

What's the procedure for landing this fix since it's sg:moderate? I couldn't find a policy about this. Do I include the bug number and summary in the commit message?  Can I land the patch and test together on mozilla-inbound?

	Matthew N. [:MattN] Assignee
	Comment 6 • 13 years ago

(In reply to Matthew N. [:MattN] from comment #5)
Jesse said to land the fix now and test separately later.

Fix landed on inbound:
https://hg.mozilla.org/integration/mozilla-inbound/rev/c244aa48ba1b

When is appropriate to land the test?

	Jonathan Kew [:jfkthame]
	Comment 7 • 13 years ago

Merged to m-c:
https://hg.mozilla.org/mozilla-central/rev/c244aa48ba1b

	Matthew N. [:MattN] Assignee
	Comment 8 • 13 years ago

Tests: https://hg.mozilla.org/integration/mozilla-inbound/rev/78e4144065cc

	Jesse Ruderman
	Comment 9 • 13 years ago

Tests: https://hg.mozilla.org/mozilla-central/rev/78e4144065cc