how to handle authenticated && !req.isAuthenticated()? about connect-auth

[ciaranj]

i'll take a look, incidentally inside your authentication provider implementation 'this.trace' should be available to you rather than console.log which should provide clearer diagnostics to you.

[ciaranj]

In the version of your code that I'm looking at, verifyAuthSuccess appears to always call 'self.success' is that intended?

[coolaj86]

Yes and no. As I'm building this out I'm adding more features.

Another bug I'm experiencing is that no matter whether I choose 'allow' or 'deny', I'm always getting a failure (which is ironic because I'm always returning success).

But since I have a failure, I figured why not handle the failure case and then I'll look into why the success case isn't working.

[ciaranj]

And when you choose allow/deny that information is being passed back in the
callback to your consumer?

On Thu, Mar 21, 2013 at 4:59 PM, AJ ONeal [email protected] wrote:

Yes and no. As I'm building this out I'm adding more features.

Another bug I'm experiencing is that no matter whether I choose 'allow' or
'deny', I'm always getting a failure (which is ironic because I'm always
returning success).

But since I have a failure, I figured why not handle the failure case and
then I'll look into why the success case isn't working.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/120#issuecomment-15251332
.

[coolaj86]

I'm not sure. I need to go back to the original example and see if the allow/deny fields are hard-coded or if I messed something up as I was editing it. I don't recall ever changing it.

Would you care to chat over skype sometime today? I'm coolaj86, I've added you.

[ciaranj]

I'm afraid I'm a bit tied up / busy atm. Is the code you've committed
up-to-date as I don't seem to have any of the login boxes etc. that your
video showed :(

On Thu, Mar 21, 2013 at 5:10 PM, AJ ONeal [email protected] wrote:

I'm not sure. I need to go back to the original example and see if the
allow/deny fields are hard-coded or if I messed something up as I was
editing my example. I don't recall ever changing it.

Would you care to chat over skype sometime today? I'm coolaj86, I've added
you.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/120#issuecomment-15252010
.

[coolaj86]

Yes, I've been testing in Chrome. There may be a bug that affects firefox.

I'll make another video in about an hour and show from git clone to up-and-running what I've done.

[coolaj86]

Did you rerun grunt build?

[ciaranj]

ok, thank you that would allow me to rather more quickly diagnose where it
might be going awry. Fwiw if req.isAuthenticated() is returning true that
means that there is a 'user' object currently stored in that session?

On Thu, Mar 21, 2013 at 5:23 PM, AJ ONeal [email protected] wrote:

Yes, I've been testing in Chrome. There may be a bug that affects firefox.

I'll make another video in about an hour and show from git clone to
up-and-running what I've done.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/120#issuecomment-15252811
.

[ciaranj]

nope, never ran that at all!

On Thu, Mar 21, 2013 at 5:23 PM, AJ ONeal [email protected] wrote:

Did you rerun grunt build?

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/120#issuecomment-15252838
.

[coolaj86]

I updated the README with the grunt build steps

[ciaranj]

You may also want to include some of the steps in here:
http://stackoverflow.com/questions/15511677/installing-npm-module-results-in-command-not-foundfor
those of us not familiar with grunt ;)

On Thu, Mar 21, 2013 at 5:25 PM, AJ ONeal [email protected] wrote:

I updated the README with the grunt build steps

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/120#issuecomment-15252988
.

[coolaj86]

if you git pull the README should have that in there

[ciaranj]

I also needed to

cd server/lib

ln -s ../../blogthing-consumer/server.js consumer.js

ln -s ../../bookface-provider/server.js provider.js

before 'node bin/demo' would work for me.

;) (sorry)

On Thu, Mar 21, 2013 at 5:35 PM, AJ ONeal [email protected] wrote:

if you git pull the README should have that in there

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/120#issuecomment-15253634
.

[coolaj86]

Try deleting and re-cloning the project. Something is out of sync with what you have and what I have.

[ciaranj]

Cloning has improved things (still got no web forms, but the sym links were
not required) ... incidentally some of the problem could be due to:

req.authenticate(['foo'], { scope: ["email", "birthday"] },

logAuthentication);

That will be having weird affects with the currently packaged version of
connect-auth. 'scope' in this context means something to connect-auth
(something not very well used or implemented I might add :( )

Try req.authenticate(['foo'], logAuthentication); isntead?

• Cj.

On Thu, Mar 21, 2013 at 5:40 PM, AJ ONeal [email protected] wrote:

Try deleting and re-cloning the project. Something is out of sync with
what you have and what I have.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/120#issuecomment-15253960
.

[coolaj86]

You were exactly right about that scope option.

Now allow and deny have the proper action.

When I click deny and then go back to click login again (from the consumer) it pops open the dialog, but skips the oauth process. However, clicking on the login (consumer side) a few more times it will eventually get there.

I think it's a cache bug where the browser assumes the previous redirect will be the current redirect. Perhaps there's a way to expire the redirect immediately so that this doesn't happen.

[ciaranj]

Possibly, you should be able to confirm that with wireshark (see if the
HTTP traffic is really leaving the browser or not), if that is an issue you
could always just append a 'cache breaker' to your url to force the browser
to always try.

On Thu, Mar 21, 2013 at 7:48 PM, AJ ONeal [email protected] wrote:

You were exactly right about that scope option.

Now allow and deny have the proper action.

When I click deny and then go back to click login again (from the
consumer) it pops open the dialog, but skips the oauth process. However,
clicking on the login (consumer side) a few more times it will eventually
get there.

I think it's a cache bug where the browser assumes the previous redirect
will be the current redirect. Perhaps there's a way to expire the redirect
immediately so that this doesn't happen.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/120#issuecomment-15261178
.