javadoc is generating "dangling" HTML links for "Miranda" methods. Either the
links should be corrected (if that is even possible) or Miranda methods should
not be linked at all.

Here's an e-mail message summarizing my investigations into this anomoly:

Date: Wed, 29 Jan 1997 14:26:30 -0800
From: msj@doppio (mark Scott Johnson [TEMP])
To: dkramer@doppio
Subject: Re: javadoc weirdness

Doug,

Here is Frank's reply to the inquiry included below.

> This is a "Miranda" method.
>
> See 8.1.2.1. If a (necessarily abstract) class implements an interface,
> but doesn't provide a definition for it, it is automatically given an
> "abstract method" which defines it.

If I'm reading his reply correctly, the behavior I noted in javadoc is
correct. However, there may still be a problem with javadoc in that the
link it generates is dangling. When I do a full rebuild of the sundocs,
I'll verify that this is the case. If so, we need to get this added to the
javadoc "to do" list.

M.S.

----- Begin Included Message -----

Date: Tue, 28 Jan 1997 09:02:13 -0800
From: msj@doppio (mark Scott Johnson [TEMP])
To: fy@doppio
Subject: Re: javadoc weirdness

Frank,

Can you shed some light on this situation? (I'd ask Mark Reinhold, but he
seems to be upset with me right now over some changes I made to the doc
comments in the java.io package. So I'm trying to stay below his radar
for awhile.) Thanks!

M.S.

----- Begin Included Message -----

Date: Mon, 27 Jan 1997 15:55:31 -0800
From: dkramer@doppio (Doug Kramer)
To: msj@doppio
Subject: Re: javadoc weirdness

I can't say with confidence what should be going on here.
If SocketOptions is a private interface, then none of its
methods should be visible, so my guess is that setOptions should not appear.
You could confirm with Frank Yellin (or Mark Reinhold).

I'd like the javadoc man page to document precisely what it
is supposed to document, so if you find special twists, please
add them to the javadoc man page.

-Doug

> Date: Sat, 25 Jan 1997 15:59:05 -0800
> From: msj@doppio (mark Scott Johnson [TEMP])
> To: dkramer@doppio
> Subject: javadoc weirdness
>
> Doug,
>
> When running the 1.1 javadoc on the java.net package, I discovered one
> strangeness: For the class java.net.SocketImpl, javadoc lists in the
> Method Index a method named "setOption", which is not defined in the
> source code for SocketImpl.java. It seems SocketImpl is an abstract class
> that implements the SocketOptions interface, which *does* implement the
> method setOption. So it seems javadoc is smart enough to figure this out
> and add setOption to the Method Index for SocketImpl.
>
> But SocketOptions is a private interface, so javadoc does not index it.
> Therefore the link for "setOption" is dangling.
>
> It seems this is either a bug or a feature of javadoc. Do you know what is
> the expected behavior and what should be done to fix the dangling link?
>
> M.S.

----- End Included Message -----

----- End Included Message -----

-- Mark Scott Johnson (impersonating Doug Kramer)