Re: [cp-patches] [Fwd: Re: KerberosPrincipal class]

[Casey Marshall]

On Nov 19, 2005, at 3:53 PM, Jeff Bailey wrote:

> On sam, 2005-11-19 at 10:20 +0100, Mark Wielaard wrote:
> > Please sent it to classpath-patches with the comments
> > belows (plus any correcttion if you want) and say that your  
> > paperwork is
> > on file with the FSF (I checked). And please mention were you  
> > would like
> > help/people to jump in.
>
> Thanks, here goes. =)
>
> 2005-11-18  Jeff Bailey  <address@hidden>
>
>         * javax/security/auth/kerberos/KerberosPrincipal.java: New  
> file.
>           (Thanks to Mark Wielaard for reviewing this!)
>
> > > There are three FIXMEs in here.  Two of them have to do with the
> > > StringTokenizer bits which I haven't tested because I'd like to make
> > > sure that the tests for these go into a testsuite somewhere,  
> > > rather than
> > > me just testing them once.
> >
> > See http://www.sourceware.org/mauve/
>
> Thanks, I'll follow up with this.
>
> > > The third one is that I'm not parsing /etc/krb5.conf right now.   
> > > Part of
> > > what I was thinking is that there should probably really be three
> > > implementations of the Kerberos stuff.  The first two are glue code
> > > around MIT Kerberos and Heimdal Kerberos.  That way if you have a  
> > > user
> > > who's already got a ticket, it continues to be useful in their Java
> > > environment.  The third is a native implementation so that people  
> > > don't
> > > need to install Kerberos in order to have classpath installed.  I
> > > imagine for now, it just means another command line switch to  
> > > configure.
> > >
> > > Does this idea work for you?  If yes, I'll slowly chew through  
> > > the rest
> > > of this over the next little bit.
> >
> > Yes I think so. You might want to CC Casey Marshall <address@hidden> who
> > knows most about the rest of our security framework.

The idea seems sound to me. The only issue I'm concerned about is  
having three implementations of the same thing, which sounds like a  
nightmare for maintenance. True, two will be glue code, which may be  
simple; it would be nice if the two glue-code layers shared as much  
code as possible.

> Cool.  I'll do that in a separate thread.  It doesn't affect this  
> class
> so much.
>
> Tks,
> Jeff Bailey
>
> <KerberosPrincipal.java>

This looks fine for the most part, since it is only a start. I would  
suggest fleshing out the Javadocs a little more, however. You should  
have at least some comments for the class (even if it's just a one- 
liner), along with @since and @author tags. The methods also need  
@param, @return, and @throws tags where appropriate (this is core  
library code, and the comments are used to build API documentation).