Newsgroups: comp.lang.dylan
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!gatech!usenet.eel.ufl.edu!news-res.gsl.net!news.gsl.net!news.mathworks.com!newsfeed.internetmci.com!in2.uu.net!demos!pluscom!usenet
From: moroz@inist.ru (Oleg Moroz)
Subject: Re: Any news from Harlequin or CMU?
X-Newsreader: Forte Agent .99d/32.182
Sender: usenet@news.rinet.ru (Ivan Frolcov)
Nntp-Posting-Host: inist.cronyx.ru
Organization: A/O Plus & Cronyx Ltd. Internet Node
Message-ID: <320ba483.10456782@news-win.rinet.ru>
References: <4tqip7$naa@cantaloupe.srv.cs.cmu.edu> <neves-0508961417240001@news.acns.nwu.edu> <4u897t$6pu@cantaloupe.srv.cs.cmu.edu> <Dvs5BC.D95@aplcenmp.apl.jhu.edu>
Date: Fri, 9 Aug 1996 20:55:52 GMT
Lines: 24

On Wed, 7 Aug 1996 18:14:00 GMT, hall@aplcenmp.apl.jhu.edu (Marty Hall) wrote:

>In article <4u897t$6pu@cantaloupe.srv.cs.cmu.edu> sef@CS.CMU.EDU
>(Scott Fahlman) writes: 
>>However, if all you have is the Java
>>byte-codes and not the sources, it will probably require some kind of
>>IDL to describe the interfaces.  C++ has the same problem.
>
>I'd be interested in what features of the Java source are hard to
>recover from the bytecodes. I have no experience with this myself, but
>the existence of free tools like "Mocha" which do an amazingly
>accurate job of recreating the source from bytecodes (to many author's
>consternation :-) made me think you could go a long way.

Well, the correspondence between .class file format and, say, Java parse tree
(with comments and layout left out) is pretty good, so that in this exact case
(interfacing _to_ classes compiled from Java) you probably don't need any
additional info besides the class file itself. But if you want to compile some
other language into Java bytecode (such as Dylan, Oberon or Modula-3) you'll
probably need some sort of program library / symbol files carrying additional
information about exported interfaces because of semantic gap between Java VM
and the corresponding language.

Oleg
