Glad to hear you're building in MAGNET support!

I don't quite understand your concern. I think the best way to achieve your
goal of providing multiple hash identifiers would be via the 'xs' "exact
substitute" parameter (or the 'as' "acceptable substitute" parameter).
See this message for details:

http://groups.yahoo.com/group/magnet-uri/message/9

I would expect software should tolerate multiple 'xt' parameters, though
they create a quandary. Considering your example,


magnet:?xt=urn:md5:94691d3ca097538863e694dd4c199359&xt=urn:sha1:4e905b1911663d35\
7cef5da8e7106c1f9ebfd856

...what if the MAGNET-handling software knows of two files, each of which
satisfies one of the hashes, but not the other? Which, then, is the
preferred item?

Use of 'xt' and either 'xs' or 'as' provides more guidance: the exact topic
always takes precedence.

Is there another aspect of your proposal I'm missing?

- Gordon @ Bitzi

PS: It seems you've written your SHA1 value in hex, rather than the Base32
that's
preferred at Bitzi and in the Gnutella (and MP2P?) networks.

Josh Guilfoyle wrote:
> It would seem that the current specification for MAGNET-URI's does not allow
> referencing single resources via multiple identifiers or names? The most
> common and useful example is to specify several hashing algorithm's output for
> a given resource so that the magnet scheme implementation can decide which to
> use for locating the resource by its URI.
>
> Currently, MAGNET-URI's are in fact allowed to specify multiple resources
> using a single resource indicator, which would make any such feature to
> provide multiple identifiers for a specific resource extremely ambiguous and
> awkward to implement. I propose that the MAGNET-URI scheme be simplified such
> that separate URIs must be required to refer to separate resources to allow
for
> multiple identifier names to exist as a description for the same resource.
>
> I realize that this makes implementation much more difficult within web
> browsers and other generic viewers, but I feel I must point out that
> cooperation with other clients currently supporting otherwise completely
> unrelated schemes is not a concern for MAGNET. Tying the specification with
> these clients may hinder future development of the specification and even
> adoption as a formal standard.
>
> For the consideration of this mailing list I have provided the following
> illustrated example:
>
>
magnet:?xt=urn:md5:94691d3ca097538863e694dd4c199359&xt=urn:sha1:4e905b1911663d35\
7cef5da8e7106c1f9ebfd856
>
>
> P.S. For those interested, my implementation of the MAGNET-URI scheme is in
> the form of a command-line utility which uses the giFT peer-to-peer framework
> (http://www.giftproject.org/) for satisfying requests. The not-yet-complete
> script may be found at:
>
> http://jasta.gotlinux.org/gift/gish-uri
>
>