Re: Revisiting Authoritative Metadata (was: The failure of Appendix C as a transition technique)

On Mon, Mar 4, 2013 at 12:50 AM, Bjoern Hoehrmann <derhoermi@gmx.net> wrote:
> I would not want to write "#encoding=windows-1252" every time I make a
> new text file, nor would I want to be bothered by a text editor add that
> for me every time (and possibly hide it so I end up forgetting it when I
> run the file through `grep` or other tools later); with that historical
> perspective the Content-Type header is a considerable improvement.
>
> (These problems suggest to me that you would rather put "identifying"
> information in the filesystem, and as a consequence convey it through
> network protocols at a higher level than "file contents", meaning you'd
> end up with something like Content-Type even under ideal circumstances.)

If it is was too much work to add it to begin with, how does it
magically end up labeled later when you scp it? This seems like a
rather spurious line of reasoning.


Of course, we cannot move away from Content-Type at this point. We're
stuck in between sniffing and having to trust Content-Type in a number
of cases for security. For new formats though such as WebVTT sniffing
for a file identifier seems to become the norm as a) it's much easier
to develop for and b) it's at least as robust as Content-Type.


-- 
http://annevankesteren.nl/

Received on Monday, 4 March 2013 10:33:20 UTC