[NUT-devel] [nut]: r644 - docs/nut.txt

[Conrad Parker]

On 01/03/2008, Michael Niedermayer <michaelni at gmx.at> wrote:
> On Fri, Feb 29, 2008 at 02:09:12PM -0500, Rich Felker wrote:
>  > On Fri, Feb 29, 2008 at 12:58:48PM +0100, Michael Niedermayer wrote:
>  >
>  > Yes, but probably in all other containers it's a single Author key
>  > with a value such as "Michael Niedermayer, Rich Felker, ..."
>  >
>  > This could also apply to disposition, but potentially makes using it
>  > harder for applications since they need to parse a sort of string list
>  > rather than just matching exact strings.
>
> There are also problems with info packet repeation if there are several
>  author tags. (add to list vs. repeated packet / already in the list)
>
>  > On the other hand, multiple values for the same key likely does not
>  > fit well with an expected API of being able to query a key and get a
>  > value as the result.

The way it works in email and HTTP headers is that multiple instances
of a key are equivalent to one instance with the values concatenated,
separated by commas.

ie.:

Author: Peter, Paul
Author: Mary

is equivalent to the single header:

Author: Peter, Paul, Mary

This allows an implementation to compare using a canonical
representation of the value of a key.

In HTTP, it allows each stage of content production to simply append
values, without modifying earlier header text. It is also valid for an
intermediate proxy to rewrite the headers in the canonical form. There
may be similar use cases in media production and distribution, like
where multiple departments in a studio add and process metadata.

cheers,

Conrad.