On 2010-06-13, at 00:20, email@example.com wrote:
> On Sun, Jun 13, 2010 at 12:09:12AM +0100, Steve Harris wrote:
Well, their hardly ambiguous. I would imagine that misleading-ness is somewhat dependent on your context.
If you don't feel comfortable with dereferencable symbols you could use schemes such as URN (http://en.wikipedia.org/wiki/Uniform_Resource_Name) which cannot be resolved, but it's missing some of the potential advantage.
There are times when having the namespace be dereferencable is an advantage. If someone discovers a LV2 turtle file in the wild, but has no idea what LV2 is (this has happened) they can paste a URI into their web browser and discover more about it. I guess it's little different to googling the fingerprint of a binary data file you find, but it's much more reliable.
The remaining advantage is really just namespacing, but using an existing globally deployed, cheap, and well understood allocation scheme - the DNS system and HTTP paths.
> And since I'm now talking to one of the experts:
You have to provide the minimum that is specified in the C header file, but you can step a long way outside it if you chose to. For example the dynamic ports extension adds a completely new type of port. I cannot be understood by hosts that do not support it of course, but they can tell that it's a plugin that they cannot make use of.
There are also extensions that are back-compatibe, and the host can identify them too.
Linux-audio-dev mailing list