[m-dev.] for review: cleanup of type_ctor_infos, relative diff 1

[Zoltan Somogyi]

On 08-Mar-2000, Fergus Henderson <fjh at cs.mu.OZ.AU> wrote:
> > +:- type module_sub_info --->
> > +	module_sub(
> 
> That change does not match the layout conventions documented
> in the "Layout" section of our Mercury coding standard.

No, but it does avoid several 

> > diff -rub --exclude CVS ws15.base/runtime/mercury_deep_copy_body.h /home/ender2/zs/ws15/runtime/mercury_deep_copy_body.h
> ...
> > @@ -513,9 +556,14 @@
> >      case MR_TYPECTOR_REP_STRING:
> >          {
> >              Word    *data_value;
> > -            int     data_tag; 
> > +			Word	data_tag;
> > +
> > +			/*
> > +			** Not all Mercury strings are aligned; in particular,
> > +			** string constants containing the empty string may be
> > +			** allocated unaligned storage by the C compiler.
> > +			*/
> >  
> > -            /* XXX simplify: tag should be zero */
> >              data_tag = MR_tag(data);
> >              data_value = (Word *) MR_body(data, data_tag);
> 
> Hmm, the code here doesn't make sense in light of the comment.
> If the pointer is not aligned, won't stripping off the tag
> give the wrong result?

The following code was not using either data_tag or data_value, except
to test whether data_value was in range; I therefore deleted them and
modified the test to test the variable data directly instead.

> It's still not clear to me why the layout_init alternative is needed.

Because unions in C do not allow you to say which member you are initializing,
and force you to initialize the first member. I want to make sure that the
cast to the first member will work properly.
--------------------------------------------------------------------------
mercury-developers mailing list
Post messages to:       mercury-developers at cs.mu.oz.au
Administrative Queries: owner-mercury-developers at cs.mu.oz.au
Subscriptions:          mercury-developers-request at cs.mu.oz.au
--------------------------------------------------------------------------