The rt command line client was just choking on the
HTTP 302 redirect (to pubcookie for the external auth).
What I ended up doing was creating a clone of the
apache virtual host config for the rt tree with a
non-pubcookied version with access restricted to 127.0.0.1.
So, is this actually a reasonable approach or have
I somehow overlooked the right way to handle this?
(RT3.8.1 on NetBSD 3.1 if it matters)
Thanks,
John Hascall
Network Infrastructure
Iowa State Univ
John Hascall wrote:
The rt command line client was just choking on the
HTTP 302 redirect (to pubcookie for the external auth).
What I ended up doing was creating a clone of the
apache virtual host config for the rt tree with a
non-pubcookied version with access restricted to 127.0.0.1.So, is this actually a reasonable approach or have
I somehow overlooked the right way to handle this?(RT3.8.1 on NetBSD 3.1 if it matters)
Better to fix rt CLI to handle external auth cases, although I’m not sure
how it wouldn’t. Authentication should be transparent to a REST client, yes?
For your solution you have to be on the RT system to use rt CLI, which isn’t
necessarily always desirable. If we actually used external auth here, I’m
sure I would have submitted a patch by now. 
– ============================
Tom Lahti
BIT Statement LLC
(425)251-0833 x 117
http://www.bitstatement.net/
– ============================
John Hascall wrote:
The rt command line client was just choking on the
HTTP 302 redirect (to pubcookie for the external auth).
What I ended up doing was creating a clone of the
apache virtual host config for the rt tree with a
non-pubcookied version with access restricted to 127.0.0.1.So, is this actually a reasonable approach or have
I somehow overlooked the right way to handle this?(RT3.8.1 on NetBSD 3.1 if it matters)
Better to fix rt CLI to handle external auth cases, although I’m not sure
how it wouldn’t. Authentication should be transparent to a REST client, yes?
I can’t think of any decent way to do this portably.
Sure, the CLI could recognize the 302-redirect,
refetch, handle the cookie business, etc,
but when it finally gets to the login form,
it would have to parse it to figure out what
form variables get the username and password
and which button is the login and so on.
For your solution you have to be on the RT system to use rt CLI, which isn’t
necessarily always desirable. If we actually used external auth here, I’m
sure I would have submitted a patch by now.
Being on the RT server is not an issue for us,
but as a general case, I see your point.
John