- From: <bugzilla@soe.ucsc.edu>
- Date: Mon, 12 Dec 2005 12:38:15 -0800
- To: w3c-dist-auth@w3.org
http://ietf.cse.ucsc.edu:8080/bugzilla/show_bug.cgi?id=112 ------- Additional Comments From julian.reschke@greenbytes.de 2005-12-12 12:38 ------- > The rationale for explicitly stating that 3xx redirects apply is as follows. > > Most use of HTTP is for reading resources. For reads, it's pretty clear what a redirect means -- go > someplace else and read what's there. > > If you have been strongly conditioned to view redirects as just applying to reads, then having redirects > apply to writes seems a bit strange. Most people have *never* worked with a system that can possibly > store your data under a different location than the one you initially specified. For example, file systems > do not work this way. As a result, even though it is a straightforward extension of the semantics of the > 3xx responses, it still runs counter to the predominant set of experiences that most programmers have. > It also seems potentially dangerous -- what if my work is written someplace I don't want it to go? Then your user agent is broken. HTTP is very clear about the fact that clients must not automatically follow redirects upon invocations of unsafe methods (see <http://greenbytes.de/tech/webdav/rfc2616.html#rfc.section.10.3> and <http://skrb.org/ietf/http_errata.html#saferedirect>). > Since this behavior is non-typical (as compared to the "typical" filesystem behavior), it makes sense for > all write operations to explicitly state that they can be redirected via 3xx. The reason we don't > exhaustively list that all HTTP responses potentially apply to all WebDAV methods is because most > HTTP response codes don't involve semantics that run counter to the typical experience of developers. > That is, 3xx is exceptional, exactly because the semantics are atypical. So again, why not state it for LOCK, PUT, PROPPATCH then? Still not convinced. ------- You are receiving this mail because: ------- You are the QA contact for the bug, or are watching the QA contact.
Received on Monday, 12 December 2005 20:38:26 UTC