Freenet compilation errors.

[Mark J. Roberts]

On Sun, 7 Jan 2001, Bryce McKinlay wrote:

> "Mark J. Roberts" wrote:
> 
> > Freenet/Core.java: In class `Freenet.Conduit':
> > Freenet/Core.java: In method `run()':
> > Freenet/Core.java:0: bad value constant type 0, index 0
> > Please submit a full bug report.
> >
> 
> I've get this quite often too. IIRC, its related to the bytecode parser. Are
> you building from pure source code, or is gcj seeing class files as well? If
> so, what did you build those class files with? There might be a PR about this
> already.

Yep, I fed gcj a list of .java files and there were already .class files
for each, so (unintuitively) it tried building the .class files instead.

My javac doesn't have a -version flag, but it's the one that came with:

java version "1.3.0"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.3.0)
Classic VM (build 1.3.0, J2RE 1.3.0 IBM build cx130-20000623 (JIT enabled: jitc))

I cleaned out the .class files, omitted the unsupported AWT code, and
specified the Latin-1 file encoding. The new log is attached. I'm asking
our local Java Gods what's up with the three super()/this errors. And the
error about the use of a sun.io.* method is probably because libjava
doesn't support special Sun libraries (although IIRC Kaffe can compile it).

So I'm down to two really evil errors:

.././Core.java:132: Internal error: Segmentation fault.
Please submit a full bug report.
 See <URL: http://www.gnu.org/software/gcc/bugs.html > for instructions.

(132 is: Key.addKeyType(Freenet.keys.KHK.keyNumber,Freenet.keys.KHK.class);)

and

../contrib/fproxy/HttpHandlerServlet.java:242: Internal compiler error in expand_expr, at expr.c:6834
Please submit a full bug report.
 See <URL: http://www.gnu.org/software/gcc/bugs.html > for instructions.

(242 is: pw.println(((FilterException) e).explanation);) 

> > .././Core.java: In class `Freenet.Core':
> > .././Core.java: In method `init(Freenet.Params)':
> > .././Core.java:132: Internal error: Segmentation fault.
> > Please submit a full bug report.
> >
> 
> Compiler bug. Its very useful if you can isolate a small test case for
> problems like this. Figuring out the test case often gets us most of the way
> towards solving the problem.

Core.java also segfaults when compiled from source (the error is formatted
differently). I'll look into it the best I can.

> > ../node/FileData.java: In class `Freenet.node.FileData$FDataInputStream':
> > ../node/FileData.java: In method `Freenet.node.FileData$FDataInputStream(Freenet.node.FileData)':
> > ../node/FileData.java:33: Can't reference `this' before the superclass constructor has been called.
> >             super(new BufferedInputStream(new FileInputStream(FileData.this.name)));
> >                                                                  ^
> > ../node/FileDataProperties.java: In class `Freenet.node.FileDataProperties':
> > ../node/FileDataProperties.java: In method `store(Freenet.FieldSet)':
> > ../node/FileDataProperties.java:81: Can't find constructor `Freenet.node.DataProperties$WrongOrderException()' in type `Freenet.node.DataProperties$WrongOrderException'.
> >             throw new DataProperties.WrongOrderException();
> >
> 
> Ditto. Bugs relating to inner classes.

This doesn't happen when compiling from source.


-- 
Mark Roberts
mjr@statesmean.com