- From: Julian Reschke <julian.reschke@gmx.de>
- Date: Thu, 8 May 2003 09:47:18 +0200
- To: "Tim Bray" <tbray@textuality.com>, "Julian Reschke" <julian.reschke@gmx.de>
- Cc: <www-tag@w3.org>
> From: www-tag-request@w3.org [mailto:www-tag-request@w3.org]On Behalf Of > Tim Bray > Sent: Wednesday, May 07, 2003 8:59 PM > To: Julian Reschke > Cc: www-tag@w3.org > Subject: Re: Draft TAG finding available: Client handling of MIME > headers > > > > Julian Reschke wrote: > > > Issue #1: clients that do PUT without passing a content type -- > what's the > > recommended server behaviour upon subsequent GETs? If the server *does* > > guess, and guesses wrong, the client ends up with a wrong > content type -- > > wouldn't be preferable then not to send a content type at all > (in which case > > the client issuing the GET would need to guess)? > > I think that in this case it's reasonable for the server to guess or to > not guess based on the application needs. I don't see an architectural > issue. OK, issue #1b: if the client doesn't supply a content type (or if the server isn't able or willing to persist the content type information supplied by the client), and the server doesn't have a "best guess" (such as when the "extension" is unknown), what behaviour is preferrable: a) upon GET, send content type "application/octet-stream" or b) upon GET, send no content type at all. Note that in case b), a client would be allowed to guess itself (it may have more information than the server!). Scenario: Client has a new HTTP-based application, let's call it myCal. Client PUTs files with extension "mycf". Server doesn't have a preference for the extension "mycf", and serves the content as "application/octet-stream". According to <http://www.w3.org/2001/tag/doc/mime-respect-20030505.html>: --- Any HTTP/1.1 message containing an entity-body SHOULD include a Content-Type header field defining the media type of that body. If and only if the media type is not given by a Content-Type field, the recipient MAY attempt to guess the media type via inspection of its content and/or the name extension(s) of the URI used to identify the resource. If the media type remains unknown, the recipient SHOULD treat it as type "application/octet-stream". --- So now when the *same* client GETs the resource through a compliant user agent, dispatching to his "mycf" application will fail, because it MUST respect the content type sent by the server (saying octet-stream). Proposals: i) allow servers not to send a content-type in case they really don't know, or alternatively ii) allow user agents to ignore the content-type if they think it was guessed (i.e., if it is "application/octet-stream"). Julian > ... -- <green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760
Received on Thursday, 8 May 2003 03:47:35 UTC