This repository has been archived by the owner on Jul 24, 2023. It is now read-only.
Unable to complete OpenID login with ruby-openid 2.9.0/2.9.1 #125
Comments
|
Okay, it seems that some parts of the code is dependent on the execution order. If anyone has the time to look into this and make a proposal for how to fix it, it would be much appreciated. I might take a look at it, but I unfortunately have very limit time available. |
Merged
|
Hi @madsolar8582, |
|
It’s the same, we are unable to complete login. |
rbebersole
added a commit
to rbebersole/ruby-openid
that referenced
this issue
May 27, 2020
|
This issue also affects one of our applications. #128 solves the problem. Any plans to merge it and release a new version? |
|
Hi @tobiashm: Are you planning to fix this issue? Thanks! |
|
This repo is being archived. Closing issue. |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
We've been testing the latest version of ruby-openid in our development environment and found that we are no longer able to successfully complete login. We started seeing
in our logs, which lead us to find out that #121 broke our login flow (note endpoint is
nil).Here is what we are doing:
OpenID::OpenIDServiceEndpoint).OpenID::Consumerto generate aOpenID::Consumer::CheckIDRequest(without discovery).OpenID::OAuth::Request) extension to the request.completemethod on the consumer.As it turns out, the call to
complete, which then calls,handle_idres, which callsid_resand causes the problem as the verification makes assumptions that are no longer true. Whenverify_discovery_resultswas beforecheck_signature, the@endpointwas set or it would perform discovery and set it. However, now thatcheck_signatureis getting called first, the@storeis nil, so theassocis set to nil and that triggers a call tocheck_auth. But,check_authwill never succeed because themake_kv_postmethod has no endpoint to call, thus preventing successful response validation and preventing login.We were able to work around this by locking down to 2.8.0. I do realize that the change made in #121 was for security reasons, but we're not sure how to proceed.
The text was updated successfully, but these errors were encountered: