Added support for other authentication mechanisms in jingle. This will allow ...

Added support for other authentication mechanisms in jingle. This will allow us to use OAuth2 tokens when OAuth2 support is ready in Talk.

BUG=None
TEST=Test cloud print and sync for regressions.


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=77434

[akalin, 2011-03-08 22:55:37]

http://codereview.chromium.org/6649006/diff/4002/jingle/notifier/base/notifie...
File jingle/notifier/base/notifier_options.h (right):

http://codereview.chromium.org/6649006/diff/4002/jingle/notifier/base/notifie...
jingle/notifier/base/notifier_options.h:44: std::string auth_mechanism;
What do you think about using an enum to pass around instead?  It's a bit more
code (since you have to have an enum to string map somewhere) but it's clearer
in client code.  Also avoids empty-string-as-default.

[sanjeevr, 2011-03-08 23:01:41]

http://codereview.chromium.org/6649006/diff/4002/jingle/notifier/base/notifie...
File jingle/notifier/base/notifier_options.h (right):

http://codereview.chromium.org/6649006/diff/4002/jingle/notifier/base/notifie...
jingle/notifier/base/notifier_options.h:44: std::string auth_mechanism;
On 2011/03/08 22:55:37, akalin wrote:
> What do you think about using an enum to pass around instead?  It's a bit more
> code (since you have to have an enum to string map somewhere) but it's clearer
> in client code.  Also avoids empty-string-as-default.

I used a string because it maps well to the SaslCookieMechanism class which also
takes in a string. That way, there are no new code changes needed to support a
new auth mechanism.

[akalin, 2011-03-08 23:44:35]

LGTM

http://codereview.chromium.org/6649006/diff/4002/jingle/notifier/base/notifie...
File jingle/notifier/base/notifier_options.h (right):

http://codereview.chromium.org/6649006/diff/4002/jingle/notifier/base/notifie...
jingle/notifier/base/notifier_options.h:44: std::string auth_mechanism;
Fair enough.  Although I expect we'll eventually standardize on oauth
eventually.

On 2011/03/08 23:01:41, sanjeevr wrote:
> On 2011/03/08 22:55:37, akalin wrote:
> > What do you think about using an enum to pass around instead?  It's a bit
more
> > code (since you have to have an enum to string map somewhere) but it's
clearer
> > in client code.  Also avoids empty-string-as-default.
> 
> I used a string because it maps well to the SaslCookieMechanism class which
also
> takes in a string. That way, there are no new code changes needed to support a
> new auth mechanism.

[akalin, 2011-03-08 23:45:56]

Actually, one more thing

http://codereview.chromium.org/6649006/diff/4002/jingle/notifier/communicator...
File jingle/notifier/communicator/gaia_token_pre_xmpp_auth.cc (right):

http://codereview.chromium.org/6649006/diff/4002/jingle/notifier/communicator...
jingle/notifier/communicator/gaia_token_pre_xmpp_auth.cc:61: // For backward
compatibility, default to kGaiaAuthMechanism if a mechanism
Hmm, I'd rather expose the default gaia auth mechanism as a variable and have
clients explicitly set that

[sanjeevr, 2011-03-09 00:54:44]

Made the changes, PTAL.

http://codereview.chromium.org/6649006/diff/4002/jingle/notifier/communicator...
File jingle/notifier/communicator/gaia_token_pre_xmpp_auth.cc (right):

http://codereview.chromium.org/6649006/diff/4002/jingle/notifier/communicator...
jingle/notifier/communicator/gaia_token_pre_xmpp_auth.cc:61: // For backward
compatibility, default to kGaiaAuthMechanism if a mechanism
On 2011/03/08 23:45:56, akalin wrote:
> Hmm, I'd rather expose the default gaia auth mechanism as a variable and have
> clients explicitly set that

Done.

[akalin, 2011-03-09 01:02:41]

LGTM

http://codereview.chromium.org/6649006/diff/7003/jingle/notifier/base/notifie...
File jingle/notifier/base/notifier_options.h (right):

http://codereview.chromium.org/6649006/diff/7003/jingle/notifier/base/notifie...
jingle/notifier/base/notifier_options.h:38: // Specifies the auth mechanism to
use ("X-GOOGLE-TOKEN", "X-OAUTH2" etc),
punctuation nit: should be:

"X-OAUTH2", etc.).