Re: The robustness principle, as view by user agent implementors (Re: Working Group Last Call: draft-ietf-httpbis-content-disp-02)

Adam Barth wrote:
>
> > The file name is advisory only. Implementations removing %-encoding
> > may be unfortunate, just like implementations stripping all but the
> > first eight [a-z] characters may be unfortunate, but they are not
> > in violation of the specification, as the file name is advisory
> > only. That there may be problems using % in file names is already
> > noted in the draft. So, we can move on with the draft as it is. I
> > am glad we cleared this up.
> 
> So, your position is that the document is meaningless then?  That's
> silly.  Why bother introducing filename* then if we don't care how
> user agents actually interpret the file name?
> 

Well, that's your position.  I'd say that C-D recognizes the reality
that the sender can't definitively specify a filename, because only the
user agent knows the constraints of its underlying filesystem.  All the
sender can do is conform to MIME syntax, and all the user agent needs
to do is understand MIME syntax.  It would be folly to attempt to make
filename anything but advisory.

-Eric

Received on Sunday, 3 October 2010 21:29:51 UTC