Hi,
sajolida@??? wrote (02 Apr 2013 10:30:52 GMT) :
>> since Tails 0.17 (I believe), various kinds of drag'n'drop (e.g.
>> tabs reordering) are broken in the Tails web browser. I was naively
>> hoping to see it automagically fixed with iceweasel 17.0.5esr, but
>> unfortunately it's still broken.
>>
>> A quick search on Tor's Trac makes me think it's a well-known
>> Torbutton bug:
>> [...]
> The weird thing is that this hits us on a point release while those bugs
> have been reported for Firefox 12 about one year ago.
... and never against the TBB, so well, this is actually specific to
Tails nowadays, and my initial analysis was wrong.
On February 28th, two patches were applied:
1. commit 58bc72e "Make use of a new TBB observer to block drag+drop"
was applied against Torbutton;
2. commit 2a543b9 "Update firefox patches for bugs #8324 and #8312"
was applied against the TBB.
Since Tails 0.17.1, we've been shipping the second patch (that is, the
new d'n'd observer) in our patched iceweasel, but we've not been
shipping the first one (that is, make use of the new d'n'd observer)
in our Torbutton. No big wonder d'n'd does not work well.
So, I guess that updating our Torbutton (todo/Torbutton_1.5) will fix
this bug.
On the short run (0.17.2 final), I wonder if we could not simply drop
the 0018-Emit-observer-event-to-filter-the-Drag-Drop-url-list.patch
Torbrowser patch introduced by TBB commit 2a543b9: in the context of
Tails, the kind of proxy bypass this patch protects against is, at
worse, a stream isolation violation (in case the OS loads an online
resource with another proxy than the one configured for iceweasel).
So, do we prefer fixing drag'n'drop (with the risk of a stream
isolation violation, and possibly regressions I can't think of) or
keep the status-quo for 0.17.2? I tend to prefer the status quo, and
wait for 0.18 to ship the correct fix.
Cheers,
--
intrigeri
| GnuPG key @ https://gaffer.ptitcanardnoir.org/intrigeri/intrigeri.asc
| OTR fingerprint @ https://gaffer.ptitcanardnoir.org/intrigeri/otr.asc
