[rfc-i] draft-flanagan-rfc-framework-00 and byte order mark (BOM)

Julian Reschke julian.reschke at gmx.de
Thu Sep 11 22:52:38 PDT 2014

On 2014-09-11 21:31, Tim Bray wrote:
> Very few things about the Web work properly without Internet Media
> Types; that’s why we put so much work into them here at the IETF.  If
> you don’t have one, you’re relying on the client software sniffing
> inside the document to figure out what it is. This is a bad, insecure
> practice.  Fortunately, on almost every web server, it is easy to set up
> content types and in most cases the right thing happens by default.  For
> example, virtually every web server I’ve been near will by default serve
> files whose names end in “.txt” with Content-type: text/plain and then
> everything will work.  I was shocked when I probed with curl and you’re
> right, no Content-type header. Which I’d call broken.  (Also both files
> are now empty…).
> Anyhow, the notion that we should try to make our stuff work without
> Internet Media Types by encouraging client sniffing is really lame.

Yes, we need to serve the resources properly over HTTP. That being said, 
including a BOM makes it possible to save the resource as a file and 
open it from the local file system.

Best regards, Julian

More information about the rfc-interest mailing list