Closed SvenDowideit closed 12 years ago
I'm currently most of the way through this, but I had to admit defeat in my html only approach, and am migrating to using handelbars templates for express serverside. Should have it up today.
I'm no expert on this but I understand that it is easy to offer shortcuts to well known open-id providers like google, twitter and facebook. I think we just shove a well known address at each of these services and that service figures out what to do. We need to look up those well known addresses.
I know google works with:
https://www.google.com/accounts/o8/id
I'm not sure who all offers OpenID, or what the normal way of making this easy is.
With OpenID now working on the express server i've launched my test wiki running the express server at:
You may notice I changed my github username to match my new home on the web. It feels like it may have been an odd thing to do, but it seemed like a good idea at the time.
@nrn :) wrt handelbars - can I possibly bully you into converting the ruby server to use it too?
it will make it much easier for us to keep the different dynamic servers in sync if we re-use the same files (I used static html only because the node.js haml impls all fell over with the tiny amount of haml that fed-wiki has.
Yeah, I made attempts with haml.js and haml-coffee previously and failed hard at both. I think it should be pretty easy to use mustache or handlebars on the ruby server, and as long as we don't do anything tricky we can consume the templates with either one. I unfortunately don't know enough ruby to muck about in the ruby server without breaking everything.
One interesting thing with the mustache templates is they are usable in a great variety of frameworks on the server side, and could also be used client side if that became desirable later.
trust me on this: ruby, and the ruby server are trivial to hack on, and Ward does a good job catching newbie (like me too) issues when merging.
fed.wiki was only the second time i'd written any ruby, and may well continue to be the only place i'm forcing myself to do so.
yes - I was intending to use the mustache templates on the client side too - cos I still contend that the server processing is redundant (except for non-js use, for which I should make another proposal :))
I'll read up on these tools.
I found http://openid.net/get-an-openid/ which lists major sites sharing their logins as OpenIDs. But heck if I can find out clear documentation what people like us are suppose to do to offer a simplified login to our users.
from my experience, lack of strong developer libraries and support are one of the bigger reasons why OpenId didn't take off, and is being replaced by other approaches. (for eg, the perl support - especially openid server, is woeful)
Mozilla's BrowserId feels to me like a better system, but it'll take some time to find out if its flawed too.
If I were wishing upon a star, I'd rather we had a federated authentication system based around gpg servers, and for browsers to use gpg for auth, content validation and encryption - damn, more stuff I won't be doing :)
(I run my own php based openid server so I get to pretend I have control over my identity)
I found this list of OpenID providers with simple universal urls:
Google : https://www.google.com/accounts/o8/id
Yahoo : https://me.yahoo.com
AOL : https://www.aol.com
LiveJournal : http://www.livejournal.com/openid/server.bml
From http://qpleple.com/how-to-make-people-login-into-your-website-with-their-google-account/
50a36cf9077b1a3d1015f2f7eb90af18d2adb6d5 adds shortcuts for loging in with googe, yahoo, aol and livejournal.
should not be hard either.