aboutsummaryrefslogtreecommitdiff
path: root/doc/index/openid/discussion.mdwn
blob: 4a50fd9dd5c39876ab1bcbecb79d0c4e2b5c906b (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
# OpenID discussion

## No return_to in OpenID server

Hi, there's no return_to from a designated OpenID server page, specs requires (I think a "should" or "must", can't recall exact wording) that it redirects back to the RP, in order to complete the registration and authentication. Unless I'm missing something, and the doc is incomplete, I'd consider this a bug. I don't expect to be of much use WRT coming up with a patch, but I'm willing to test ;-) .

> If this is a bug, could you please explain:
> 
> * What happens when the bug occurs?
> * How can one reproduce the bug?
>
> PS, please file bugs under [[bugs]] in future. --[[Joey]]

>> Oops, my bad, didn't know that existed at the time I wrote this.
>>
>> What happened is that the process wouldn't complete, therefore I couldn't login with my OpenID.
>>
>> reproducibility: every time
>>
>> Should probably move this page, eh? ;)
>> I'd do that, but I dunno know other than using the SCM backend in question....

Here's some actual output (with my OpenID URL stripped out):

do=postsignin&oic.time=1238224497-1450566d93097caa707f&openid.assoc_handle=%7BHMAC-SHA1%7D%7B49cdce76%7D%7BBhuXXw%3D%3D%7D&openid.identity=|<==== MY OPENID URL GOES HERE ====>|&openid.mode=id_res&openid.op_endpoint=http%3A%2F%2Fwww.myopenid.com%2Fserver&openid.response_nonce=2009-03-28T07%3A15%3A02ZDUFmG3&openid.return_to=http%3A%2F%2Fsimonraven.kisikew.org%2Fbin%2Fikiwiki.cgi%3Fdo%3Dpostsignin%26oic.time%3D1238224497-1450566d93097caa707f&openid.sig=E51Xh6Gnjku%2B0se57qCyhHbT5QY%3D&openid.signed=assoc_handle%2Cidentity%2Cmode%2Cop_endpoint%2Cresponse_nonce%2Creturn_to%2Csigned

The `return_to` arg should NOT be `signed`, it should be the originating URL where you initially logged in.

>> Yes, exactly. That's also my understanding of the spec.

> I think you're confusing 'openid.return_to' with 'return_to'. The
> former is present above, and is, indeed, the originating url, the latter
> is part of the *value* of the 'openid.signed' parameter generated by myopenid.com. --[[Joey]] 

Also, I dunno what the assoc_handle is doing spitting out an arg like `{HMAC-SHA1}{49cdce76}{BhuXXw%3D%3D}` it should be processed further. I have the needed perl packages installed (latest for Lenny). Hrm, would endianness matter?

> Again, this value is created by the openid server, not by ikiwiki.
> I see the same HMAC-SHA1 when using myopenid, and completly different
> things for other openid servers. (Ie, when using livejournal as an openid server,
> `openid.assoc_handle=1239305290:STLS.QiU6zTZ6w2bM3ttRkdaa:e68f91b751`)

>> OK, I wasn't too sure about that, seemed bogus or somehow wrong or in error, like it wasn't actually being `completed`.

> I'm fairly sure that is all a red herring.
> 
> So, when I was talking about reproducing the bug, I was thinking perhaps you could tell me what openid server you're using,
> etc, so I can actually see the bug with my own eyes.

>> myopenid.com, with the CNAME option turned on.

> The sanitised url parameters you've provided are not generated by ikiwiki at all.
> They don't even seem to be generated by the underlying [[!cpan Net::OpenID]] library.

>> That was a server log entry with date/host/time stripped, and my URL also stripped. Everything else is as was in the log. I installed the Debian packages in Lenny, both server and consumer OpenID Perl packages.

> I'm pretty sure that what you're showing me is the url myopenid redirects
> the browser to after successfully signing in. At that point, ikiwiki
> should complete the signin. What fails at this point? How can I reproduce this failure? --[[Joey]] 

I'll try it again myself. I had tried it oh probably 6 times before I finally gave up on it. Maybe I'm getting rusty and I'm just PEBKACing all over the place. :P

Also, to address the point about this discussion being in the wrong area (not under bugs), should I move it, or will you? I don't mind doing it, if you can't.