Re: Bug? - japitools - Dropped protected method not detected

[Stuart Ballard]

Peter Connolly wrote:
> When I delete a protected method from a new version of a jar and compare
> the old vs. the new using japitools, the report does not flag the fact
> that the method is missing.
>
> According to the list of illegal changes between jar files:
>
> http://www.kaffe.org/~stuart/japi/jcompat.txt
>
>
> Methods:
> - ILLEGAL: deleting public/protected methods
>
> ...so this should have been reported, right???

Try using zgrep (or gunzip and grep separately) for the name of the 
method you dropped, in each of the two japi files. If it's correctly 
present in the first one and correctly NOT present in the second, then 
there's a problem in japicompat; otherwise, there's a problem in Japize.

Unfortunately I don't have much (any, really) time to maintain japitools 
these days. I'm more than willing to help track down the problem and to 
apply a patch to fix it, but unless it's really simple and obvious, I 
probably don't have time to write the patch myself.

Oh, and as far as finalize() is concerned: That's not an incompatible 
change, the method is always present in the superclass because it's 
inherited from java.lang.Object. Japitools (and Sun) don't require that 
the method be implemented at the same level of the inheritance hierarchy 
in subsequent versions, just that it's present somewhere in the 
superclass chain. Perhaps the same thing is happening with the other 
method you removed?

> The reason we're so suspicious is that we NEED tools like JApiTools to
> help us recognize when developers here are committing incompatible
> changes.  We're not as interested in comparing different versions of JDK
> as we are in comparing changes within our own .jar files.

I'd be really interested in having some help maintaining japitools. If 
you're using it on a regular basis, would you consider helping to 
maintain it? It's obviously not much work except for when something 
breaks - and even only doing a very small amount is an improvement on 
what I've been able to do lately :)

Since the last release I've had various reports of incorrect behavior 
that I've not been able to find time to track down. Michael Koch was 
doing some work on false-positive supression but I haven't heard from 
him lately; I never saw the unfinished patches (and certainly wouldn't 
have had time myself to finish them).

If you (or anyone else) is interested in helping to maintain japitools, 
I'd be happy to set up a project on Gna! (or Savannah if they'd only 
start accepting new projects) to coordinate the maintenance.

Thanks for your interest in japitools :)

Stuart.

--
Stuart Ballard, Senior Web Developer
NetReach, Inc.
(215) 283-2300, ext. 126
http://www.netreach.com/