Add an explicit way of forcing the MDnsClient to listen on the network

Add an explicit way of forcing the MDnsClient to listen on the network

Add an object called MDnsListenRef, whose existence forces the MDnsClient to
continue listening on the network, but who has no other function.

Rationale: Since a sandboxed process using an MDnsClient may not be able
to create new sockets on demand, it makes sense to force the MDnsClient to
be awake for the lifetime of the process. Add the ability to force the
MDnsClient to stay awake.

Also useful for abstractions over MDnsClient where you know that it makes sense
not to immediately shut down the client even if we shut down all listeners and
transactions.

BUG=233821
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=209812

[szym, 2013-06-26 18:52:48]

I'm a little bit worried of feature creep. Do you anticipate you will need both
behaviors: 1) where MDnsClient stops listening when there are no listeners; 2)
where MDnsClient listens all the time?

[Noam Samuel, 2013-06-26 21:30:09]

On 2013/06/26 18:52:48, szym wrote:
> I'm a little bit worried of feature creep. Do you anticipate you will need
both
> behaviors: 1) where MDnsClient stops listening when there are no listeners; 2)
> where MDnsClient listens all the time?

Hm. Looks like the correct solution if the MDnsClient lives in a sandboxed
process will be to remove ref-counting entirely. Since the discussion hasn't yet
settled down yet, it may make sense sense to sit on this issue for a bit while
this gets worked out.

[Noam Samuel, 2013-06-29 00:19:12]

On 2013/06/26 21:30:09, Noam Samuel wrote:
> On 2013/06/26 18:52:48, szym wrote:
> > I'm a little bit worried of feature creep. Do you anticipate you will need
> both
> > behaviors: 1) where MDnsClient stops listening when there are no listeners;
2)
> > where MDnsClient listens all the time?
> 
> Hm. Looks like the correct solution if the MDnsClient lives in a sandboxed
> process will be to remove ref-counting entirely. Since the discussion hasn't
yet
> settled down yet, it may make sense sense to sit on this issue for a bit while
> this gets worked out.

OK, uploaded a new version of the CL that drops all reference counting semantics
on MDnsClientImpl in favor of explicit start/stop semantics exposed in the
MDnsClient interface.

[gene, 2013-07-02 03:39:16]

lgtm, but I'll let szym@ to take a look

nit:

https://codereview.chromium.org/17922002/diff/8001/net/dns/mdns_client.h
File net/dns/mdns_client.h (right):

https://codereview.chromium.org/17922002/diff/8001/net/dns/mdns_client.h#newc...
net/dns/mdns_client.h:149: virtual bool GetListening() const = 0;
IsListening() ?

[szym, 2013-07-02 17:27:56]

https://codereview.chromium.org/17922002/diff/8001/net/dns/mdns_client.h
File net/dns/mdns_client.h (right):

https://codereview.chromium.org/17922002/diff/8001/net/dns/mdns_client.h#newc...
net/dns/mdns_client.h:148: virtual void StopListening() = 0;
Is it safe to call this from callbacks?

https://codereview.chromium.org/17922002/diff/8001/net/dns/mdns_client.h#newc...
net/dns/mdns_client.h:149: virtual bool GetListening() const = 0;
On 2013/07/02 03:39:16, gene wrote:
> IsListening() ?

I concur it should be IsListening.

[Noam Samuel, 2013-07-02 18:08:22]

https://codereview.chromium.org/17922002/diff/8001/net/dns/mdns_client.h
File net/dns/mdns_client.h (right):

https://codereview.chromium.org/17922002/diff/8001/net/dns/mdns_client.h#newc...
net/dns/mdns_client.h:148: virtual void StopListening() = 0;
On 2013/07/02 17:27:56, szym wrote:
> Is it safe to call this from callbacks?

No, though I think in this case it makes sense to add a comment to this effect
rather than make it safe. Comment added.

https://codereview.chromium.org/17922002/diff/8001/net/dns/mdns_client.h#newc...
net/dns/mdns_client.h:149: virtual bool GetListening() const = 0;
On 2013/07/02 17:27:56, szym wrote:
> On 2013/07/02 03:39:16, gene wrote:
> > IsListening() ?
> 
> I concur it should be IsListening.

Done.

[szym, 2013-07-02 19:33:40]

lgtm

[commit-bot: I haz the power, 2013-07-02 20:22:21]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/noamsml@chromium.org/17922002/13001

[commit-bot: I haz the power, 2013-07-02 23:26:28]

Change committed as 209812