On Fri, 2010-10-08 at 11:03 -0300, muriloo at br.ibm.com wrote: > > Thanks for your reply David. > > I think we could implement keyring support for password first and after > implement a function to test if cookie is still valid and save cookie in > gnome-keyring either. Yeah, that definitely sounds like the sanest approach. > For now, I've drafted a patch to add gnome-keyring support for user's > password. Please refer to the attachment openconnect-add-gnome-keyring-support.patch Hm, that doesn't seem generic enough. Remember, the server can present you with *arbitrary* forms to fill in. Some text input boxes are marked as 'text' and some are marked as 'password' but there can be any number of each. Just because it's *common* to have a single username and a single password entry, that doesn't mean it's all we have to cope with. See how the non-password entries are saved in gconf with under a key named 'form:$AUTH_ID:$OPT_NAME'? We want something similar for password entries, I think -- so it can save *any* of the text boxes, not just one. Or am I misreading your patch? -- dwmw2