[m-rev.] for review: fix float_to_string so that roundtripping works

[Ralph Becket]

Fergus Henderson, Thursday, 21 November 2002:
> On 20-Nov-2002, Peter Ross <pro at missioncriticalit.com> wrote:
> > +++ library/string.m	20 Nov 2002 16:42:19 -0000
> > @@ -1826,45 +1826,21 @@
> >  :- pragma foreign_proc("C",
> >  	string__float_to_string(Flt::in, Str::uo),
> >  		[will_not_call_mercury, promise_pure, thread_safe], "{
> > +	char buf[ML_SPRINTF_FLOAT_BUF_SIZE];
> > +	ML_sprintf_float(buf, Flt);
> > +	MR_make_aligned_string(Str, buf);
> 
> That needs to be MR_make_aligned_string_copy().
> 
> > +:- pragma foreign_code(c, "
> > +void
> > +ML_sprintf_float(char *buf, MR_Float f)
> > +{
> > +	MR_Float round;
> > +	int 	 i = ML_MIN_PRECISION;
> > +
> > +	do {
> > +		sprintf(buf, ""%#.*g"", i, f);
> > +		if (i >= ML_MAX_PRECISION) {
> > +			/*
> > +			** This should be sufficient precision to
> > +			** round-trip any value.  Don't bother checking
> > +			** whether it can actually be round-tripped,
> > +			** since if it can't, this is a bug in the C
> > +			** implementation.
> > +			*/
> > +			break;
> > +		}
> > +		sscanf(buf, ML_FMT, &round);
> > +		i++;
> > +	} while (round != f);
> > +}
> > +").

Small point, but isn't a for loop more appropriate here?

	for(i = ML_MIN_PRECISION; i <= ML_MAX_PRECISION; i++) {
		sprintf(buf, "%#.*g", i, f);
		sscanf(buf, ML_FMT, &round);
		if(round == f) {
			break;
		}
	}

- Ralph
--------------------------------------------------------------------------
mercury-reviews mailing list
post:  mercury-reviews at cs.mu.oz.au
administrative address: owner-mercury-reviews at cs.mu.oz.au
unsubscribe: Address: mercury-reviews-request at cs.mu.oz.au Message: unsubscribe
subscribe:   Address: mercury-reviews-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------