On Wed, 25 Jan 2006, Neil Van Dyke wrote:

>Lauri Alanko <xxxxxx@iki.fi> wrote at 2006-01-25T19:19:28+0100:
>> Any sane implementation will first parse the URI into its constituents
>> and form a list of path segments, and then operate on that list. It

>Regarding this assertion, one additional data point: my "uri.scm"
>library ("http://www.neilvandyke.org/uri-scm/") supports both string and
>parsed representations of URIs, and allows them to be intermixed.  I
>expect most programmers will just use the string representations, as
>they are more convenient and familiar in general.

Right.  This is probably the best approach, in fact.  Let
individuals decide whether they want strings as URIs, and
trust in them to work it out correctly.

But if we're to leave URI's polymorphic (admitting of two
or more syntaxes/structures), does it still make sense to
use them to identify modules?

I don't, in general, like a module naming convention that
strongly hints that the modules are to be found over a
network; some very intelligent idiot is bound to think
a single code repository is a good idea and subvert
security on entire networks if a cracker can get to it.

				Bear