URIs

[Raynes]

Y U NO URI?

I wanted to use this to parse a mongodb:// URL, but java.net.URL wont have it. How can we sort this out?

[cemerick]

Well, because the name on the tin is "url", not "uri". Sorry.

Seriously though, this is just business-as-usual for (certain) JDK classes. java.net.URL will throw an exception for any URL whose protocol part does not correspond with a known "URL stream handler" (i.e. support for loading a resource from the given URL, which is what supports e.g. slurp).

As you imply, the answer to this may be to use java.net.URI instead. I considered this, but there are subtleties to URIs that are not shared by URLs, and I didn't want to get hung up on them at the time.

Q: are mongodb connection strings URLs, or URIs?

[ianmcorvidae]

I'd also love if this module used java.net.URI, but I'm also sympathetic to the complexities of non-URL URIs. My answer to the last question would be that these (or the postgresql:// ones I wanted to use!) are in fact URLs, on the basis of the RFC definition:

A URI can be further classified as a locator, a name, or both. The
term "Uniform Resource Locator" (URL) refers to the subset of URIs
that, in addition to identifying a resource, provide a means of
locating the resource by describing its primary access mechanism
(e.g., its network "location").

These sorts of URLs share the property java.net.URL-compliant ones do, that they specify an access mechanism (generally, a scheme, DNS name, port, database name in the path, userinfo). Curiously, to me, java.net.URL seems to have a different understanding of the differences:

A URI is a uniform resource identifier while a URL is a uniform resource locator. Hence every URL is a URI, abstractly speaking, but not every URI is a URL. This is because there is another subcategory of URIs, uniform resource names (URNs), which name resources but do not specify how to locate them. The mailto, news, and isbn URIs shown above are examples of URNs.

Everything else I can find seems to argue (reasonably, I think) that mailto,news,tel, etc. also specify access mechanisms and are therefore URLs, even though they are not necessarily transfer-protocol-over-TCP access mechanisms.

I do notice the cljs version uses goog.Uri, which probably has URI semantics and not java.net.URL or RFC URL semantics.

Personally, I'd like it if this used java.net.URI, and perhaps tried to convert to a URL where java will let it, but I also don't know what the downstream consequences of that would be.