Disable RegisterProtocolHandler on Linux.

Disable RegisterProtocolHandler on Linux.

The OS-level registration isn't implemented on Linux, and so every time the
browser starts it will act as though another program has been registered to
handle the protocols it knows about and deregister them all internally.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=88246

[koz (OOO until 15th September), 2011-06-06 06:42:05]

Hey Tony,

This needs to go in to the branch for M13, could you tell me what the process is
for that?

Cheers,
James

[koz (OOO until 15th September), 2011-06-06 07:01:57]

Oh, never mind - I figured it out :-)

[tony, 2011-06-06 16:48:17]

This will disable it on ChromeOS too.  Is that what we want?  Ideally on
ChromeOS it wouldn't ever unregister since there are no other apps that could
register a protocol.

In general, I think it's OK to disable RPH for Linux for M13, but please make it
a priority to get it working for M14.

[koz (OOO until 15th September), 2011-06-07 07:51:58]

Ah okay. I think it should be okay to disable on ChromeOS, as they seem to have
their own custom mailto support anyway:

http://src.chromium.org/viewvc/chrome/trunk/src/chrome/browser/platform_util_...

Ben, what do you think?

[benwells, 2011-06-07 10:24:04]

We should probably aim to turn it on in ChromeOS so we can support other mail
providers, other protocols / web apps and so we don't need any special case for
this platform.

But maybe this should wait until M14. Looking at the settings pages on ChromeOS
they seem to be a bit different, I'm not sure if what we've done there will just
drop in? (I am looking on rev 12, I haven't seen the new rph stuff on ChromeOS).
Apart from that it should just work. 

As Tony mentions not having the OS integration on ChromeOS isn't a problem, but
we definitely need to stop unregistering at startup if we do turn it on.

[benwells, 2011-06-07 11:03:15]

> Looking at the settings pages on ChromeOS
> they seem to be a bit different,

Looking again there are just extra options pages so there shouldn't be a
problem. As the ChromeOS stuff is done as an external handler it shouldn't clash
with the rph stuff (if there is an rph handler it doesn't get to external
handlers). If we get this done now for ChromeOS its one less thing to worry
about for M14.

If we turn ChromeOS on should we 'pre-register' gmail?

[laforge, 2011-06-07 19:13:39]

Time is off the essence on m13, please close this out.

On 2011/06/07 11:03:15, benwells wrote:
> > Looking at the settings pages on ChromeOS
> > they seem to be a bit different,
> 
> Looking again there are just extra options pages so there shouldn't be a
> problem. As the ChromeOS stuff is done as an external handler it shouldn't
clash
> with the rph stuff (if there is an rph handler it doesn't get to external
> handlers). If we get this done now for ChromeOS its one less thing to worry
> about for M14.
> 
> If we turn ChromeOS on should we 'pre-register' gmail?

[tony, 2011-06-07 19:19:49]

Since ChromeOS already has a way to handle mailto, this change seems OK, but
suboptimal.  LGTM.

However, I would rather we just never deregister on Linux except via the options
dialog.