On Sun, 2008-03-16 at 18:51 +0100, Pierre THIERRY wrote:
> Scribit Swaroop Sridhar dies 16/03/2008 hora 12:21:
> > A top-level form in BitC must be a defining form. That is, one of
> > define, defstruct, defunion, defrepr, deftypeclass, definstance.
> > 
> > The parser is looking for one of these keywords, to begin a top-level
> > form, which is why you are getting a syntax error after the end of the
> > first word.
> 
> I would be useful to get a more verbose error.

Yes. For a research compiler, BitC is quite good about errors after the
parse stage. Generating good errors during parsing is a fairly black
art, and Yacc/Bison aren't really very good tools for it. For a research
compiler, this wasn't a priority, but for any production implementation
it certainly should be.

> I had to manually tell GCC which standard I want:
> 
> $ gcc -O2  -I ../include -L ../lib -o bitc.out bitc.out.c -lbitc -lgc 
> --std=c99

> And it runs, too:
> 
> $ ./bitc.out
> Hello, world!
> 
> The attached one-line patch seems to fix this. The second one is a
> convenience: it adds an EOL after the output of GCC's invocation on
> stderr.

I just applied both of those.


shap

_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev

Reply via email to