When should an asterisk be encoded in an HTTP URL?

According to RFC1738, an asterisk (*) "may be used unencoded within a URL":

Thus, only alphanumerics, the special characters "$-_.+!*'(),", and reserved characters used for their reserved purposes may be used unencoded within a URL.

However, w3.org's Naming and Addressing material says that the asterisk is "reserved for use as having special signifiance within specific schemes" and implies that it should be encoded.

Also, according to RFC3986, a URL is a URI:

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").

It also specifies that the asterisk is a "sub-delim", which is part of the "reserved set" and:

URI producing applications should percent-encode data octets that correspond to characters in the reserved set unless these characters are specifically allowed by the URI scheme to represent data in that component.

It also explicitly specifies that it updates RFC1738.

I read all of this as requiring that asterisks be encoded in a URL unless they are used for a special purpose defined by the URI scheme.

Is RFC1738 the canonical reference for the HTTP URI scheme? Does it somehow exempt the asterisk from encoding, or is it obsolete in that regard due to RFC3986?

Wikipedia says that "[t]he character does not need to be percent-encoded when it has no reserved purpose." Does RFC1738 remove the reserved purpose of the asterisk?

Various resources and tools seems split on this question.

PHP's urlencode and rawurlencode-- the latter of which purports to follow RFC3986 -- do encode the asterisk.

However, JavaScript's escape and encodeURIComponent do not encode the asterisk.

And Java's URLEncoder does not encode the asterisk:

The special characters ".", "-", "*", and "_" remain the same.

Popular online tools (top two results for a Google search for "online url encoder") also do not encode the asterisk. The URL Encode and Decode Tool specifically states that "[t]he reserved characters have to be encoded only under certain circumstances." It goes on to list the asterisk and ampersand as reserved characters. It encodes the ampersand but not the asterisk.

Other similar questions in the Stack Exchange community seem to have stale, incomplete, or unconvincing answers:

urlencode() the 'asterisk' (star?) character This question highlights the differences between Java's and PHP's treatment of the asterisk and asks which is "right". The accepted answer references only RFC1738, not mentioning the more recent RFC3986 and resolving the conflict. Another answer acknowledges the discrepancy and suggests that asterisks are different for URLs specifically, as opposed to other URIs, but it doesn't provide specific authority for that conclusion.
Can an URL have an asterisk? One answer cites only the older RFC1738 and the accepted answer implies it's acceptable when being used as a delimiter, which one presumes is the "reserved purpose".
Can I use asterisks in URLs? The accepted answer seems to discourage use of the asterisk without clarifying the rules governing the use. Another answer says you can use the asterisk "because it's a reserved character". But isn't that only true if you're using it for its reserved purpose?
escaping special character in a url One answer points out that "there is some ambiguity on whether an asterisk must be encoded in a URL". I'm trying to resolve that ambiguity with this question.
Spring UriUtils and RFC3986 This question notes that UriUtil's encodeQueryParam purports to follow RFC3986, but it doesn't encode the asterisk. There are no answers to that question as of 2014-08-01 12:50 PM CDT.
How to encode a URL in JavaScript? This seems to be the canonical JavaScript URL encoding question on Stack Overflow, and although the answers note that asterisks are excluded from the various methods, they don't address whether they should be.

With all this in mind, when should an asterisk be encoded in an HTTP URL?

Solution

Short answer

The current definition of URL syntax and W3 standards indicates that you never need to percent-encode the asterisk character in the path, query, or fragment components of a URL:

* is defined as a sub-delimiter in RFC 3986 which is explicitly allowed in the path, query and fragment.
The W3 definition of URI syntax specifies !*"'(), as "extra", which are allowed in the path, query and fragment.

HTTP 1.1

RFC 3986 (current URI syntax)

Asterisk is reserved as a sub-delimiter: a character that has no special meaning in the generic URI syntax, but can be used by implementations to subdivide a component. The path, query and fragment parts of a URI all allow the set of characters defined as pchar, which includes sub-delimiters:

reserved      = gen-delims / sub-delims
sub-delims    = "!" / "$" / "&" / "'" / "(" / ")"
              / "*" / "+" / "," / ";" / "="
...
pchar         = unreserved / pct-encoded / sub-delims / ":" / "@"

Interestingly the HTTP spec doesn't specify the meaning of any additional delimiters in these URI components that you might associate with HTTP. The use of & as a delimiter for query strings is specified by W3:

4.10.22.6 URL-encoded form data

Note: This form data set encoding is in many ways an aberrant monstrosity, the result of many years of implementation accidents and compromises leading to a set of requirements necessary for interoperability, but in no way representing good design practices.

...

Let strings be the result of strictly splitting the string payload on U+0026 AMPERSAND characters (&).

RFC 2396 (URI spec before January 2005)

Obsoleted by RFC 3986 above.

* is listed as an "unreserved character" in RFC 2396, which is used to define URI syntax in HTTP 1.1. Unreserved characters are allowed in the path component of a URL.

2.3. Unreserved Characters

Data characters that are allowed in a URI but do not have a reserved purpose are called unreserved. These include upper and lower case letters, decimal digits, and a limited set of punctuation marks and symbols.
   unreserved  = alphanum | mark

   mark        = "-" | "_" | "." | "!" | "~" | "*" | "'" | "(" | ")"
Unreserved characters can be escaped without changing the semantics of the URI, but this should not be done unless the URI is being used in a context that does not allow the unescaped character to appear.

HTTP 1.0

HTTP 1.0 references RFC 1738 to define URL syntax, which through a series of updates and obsoletes means it uses the same RFC as HTTP 1.1 for URL syntax.

As far as backwards compatibility goes, RFC 1738 specifies the asterisk as an unreserved character which can be used unencoded:

Thus, only alphanumerics, the special characters "$-_.+!*'(),", and reserved characters used for their reserved purposes may be used unencoded within a URL.

This means you're safe to use an asterisk unencoded, even in URLs pointing to the oldest of systems.

As a side note, the asterisk character does have a special meaning in a Request-URI in both HTTP specs, but it's not possible to represent it with an HTTP URL:

The asterisk "*" means that the request does not apply to a particular resource, but to the server itself, and is only allowed when the method used does not necessarily apply to a resource. One example would be
   OPTIONS * HTTP/1.1

Disclaimer: I'm just reading and interpreting these RFCs myself, so I may be wrong.