What's wrong with this code? - nevermind

[Jim White]

Hmmm, that was some strangeness in updating CVS on my PowerBook's copy. 
  Updating another copy on my G5 worked fine (as did nuking the gnu 
subdir on my PowerBook and upating - apparently MacCVSClientX didn't 
believe me that I wanted to overwrite local changes [caused by 
preprocessing]).

Perhaps N H could try getting a fresh copy from CVS (or rm -rf gnu ; cvs 
update -d) if the trouble is with an updated copy, conflict messages, or 
the like.

My struggles with GCJ for native PPC continue.  I've installed Yellow 
Dog Linux 4.0.1 on one of my G5s.  Had to sort out the classpath problem 
for org.w3c.dom.Node and friends (including that Classpath.org supplies 
DOM Level 3 and Kawa only implements Level 2) because --disable-xml does 
not remove a Common Lisp dependency on that stuff.  Anyhow, once I 
stopped trying to just substitute Classpath 0.14 for GCJ 3.3 libgcj, I 
got as far as having a bunch of missing symbols amongst the Java files 
so that it won't link even though they all compiled OK.

Jim

Jim White wrote:

> Per Bothner wrote:
> 
>> N H wrote:
>>
>>> Unfortunately, right now the CVS code is not building. I got errors 
>>> compiling the scm files first in files.scm and then in misc.scm
>>
>>
>>
>> Are you sure?  Is anyone else seeing this?
> 
> 
> I'm getting:
> 
> CallContext.java:336: cannot resolve symbol
> symbol  : class FluidBinding
> location: class gnu.mapping.CallContext
>   public FluidBinding fluidBindings;
> 
> and a bunch more like it as there is no file FluidBinding.java that I 
> can find.
> 
> And while CallContext.java seems to be getting preprocessed, I don't 
> think that is the version being compiled.
> 
> Jim
> 
>