Re: [Chicken-users] are standard bindings write protected?

[Joerg F. Wittenberger]

> > It appears to be actually worse than just that.  The code, compiled in
> > module A modifies the globals, code in the module B (depending on A),
> > uses it, but gets the original definition instead my modified one.
> >
> > I also see some suspicious messages relating to other definitions I
> > wrote over.  Like here:
> >
> > Warning: global variable `abort' is never used
> > Warning: global variable `signal' is never used
> > Warning: global variable `##sys#error' is never used
> 
> You are probably compiling in block-mode, or using
> the `export' declaration, without exporting the variables mentioned
> above.

I tried to export those: the programm continues a little further
(until the next bug of mine).  Not enough to really say anything about
the effects.  Thanks.

For clarification: I assumed that I would _import_ 'abort', 'force',
'signal', '##sys#error' etc. from the 'library' unit and just 'set!'
the binding to a new value.  Should I really export those imported
bindings to get the replaced globally?  I'd say that would be counter
intuitive.

At least it raises the question (for me) how other (standard) units
will behave.  Besides 'force' I'm replacing the exception handling
mechanism and it would kill me if the standard units would not
signaling their errors into my exception handler.

> > (future-result-set! f (with-exception-guard identity thunk))
>                                         ^^^^^
>                                         handler
> 
> (the error doesn't get shown, the thread terminates, but since
> there is no join, the exception doesn't get signalled.

This at least explains the loop in the test, thanks.

/Jörg

-- 
The worst of harm may often result from the best of intentions.