looking for feedback on REDIRECT issue lc-57-noautoupdate

Hi,

see 
(<http://greenbytes.de/tech/webdav/draft-ietf-webdav-redirectref-protocol-latest.html#rfc.issue.lc-57-noautoupdate>).

Back in 2000, Yaron Goland wrote that allowing a server to automatically 
update the target of a redirect ref can lead to surprising results.

I agree with this statement, but as a matter of fact there are systems 
that actually *do* behave this way (one of which being one we 
implemented...). Also, the REDIRECT spec IMHO empower users to create, 
modify and discover HTTP redirects without putting too many constraints 
on how they actually behave once they've been created.

Back in January I wrote:

"I don't think we can forbid that. This spec consists of (a) 
clarifications of how a server that supports redirects should behave for 
specific WebDAV methods, and (b) extensions to explicitly create them 
(or to apply a method to the redirect itself). As such, we shouldn't add 
any requirements that HTTP doesn't add. What we could do is (1) note why 
auto-update may be a bad idea, and possibly (2) define that redirects 
created by MKREDIRECTREF should not behave that way (or alternatively 
define more specific resource types)."

Today, I lean towards (1), that is we add a statement that redirects 
that change their link target automatically can cause confusion, but we 
do not forbid them.

Unless somebody objects, I'll make that change in the next draft.

Best regards, Julian


-- 
<green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760

Received on Wednesday, 6 October 2004 17:39:16 UTC