Comment 61 for bug 857153

> [Approval Request Comment]
> Regression caused by (bug #):

no bug, gnome's move away from gconf

> Testing completed (on m-c, etc.):
> Risk to taking this patch (and alternatives if risky):
> The old behavior of checking gconf will be used if the dbus call does not
> succeed (ie, if AT-SPI2 is not installed), so I think that the risk is small.

The it is theoretically possible this could somehow cause a11y to be enabled by accident which would cause a perf regression, however I would evaluate the probaility of this as low, and if it was a problem I would expect it to be a fairly general problem which we should have already heard about.

> An alternative approach would be to modify the start-up script to call
> dbus-send and set GNOME_ACCESSIBILITY=1 if the call indicates that
> accessibility is enabled, but this would likely affect start-up time more
> than applying the patch would, so applying the patch seems preferable.

The shell script went away, so this might not be a possibility, and it would be completely new untested code.
The regression is pretty significant (we've seen mails from people wondering why firefox is no longer accessible), and since Fire Fox 10 will be an esr and so around for a while to some degree, I'll do my best to take Surkov's place (he's out) and say I think we want to take this.