Comment 12 for bug 663923

Per the comment from stuartm on the RT, he and I had discussed that mpopenid should still be used, because it is the only thing that currently has the desired behavior. He felt that getting the apache-openid plugin to have the desired behavior would take longer. He expressed willingness to help with getting that working. Meanwhile, though, he felt that mpopenid should be made to work as it was before. My vague thought as to resolution was this:

1) I find out where mpopenid is, and try to figure out why it isn't working.
2) LOSAs get it working, with whatever help I can provide. This bug is now fixed, yay. We probably add a bug about dropping mpopenid and switching to apache-openid.
3) ISD eventually adds the features we need in apache-openid to replace mpopenid.
4) LOSAs switch. The bug to drop mpopenid is closed, yay.

Therefore, I started trying to get someone to show me the mpopenid code. This would help with steps 1 and 3. Getting a copy of mpopenid has proven to be difficult; I do not have one yet, after asking a group that includes Barry, Chex, and Francis. However, it seems we know the answer as to why mpopenid is not working: it is gone, replaced by apache-openid before apache-openid was ready to replace it.

The plan I outlined above still seems like the right thing to do for now, assuming that it will let us resolve this issue significantly faster than trying to get apache-openid to work.

If apache-openid needs to be resolved now, ISD certainly has the expertise and is the code owner, but Foundations will step up to the plate if we have to.

Either way, I suspect that finding a copy of mpopenid will help apache-openid get the needed features. The LOSAs are our only hope of that now, AFAIK.