Re: Defect in linearization of short circuit &&

["Jacek Śliwerski" <sliwers@googlemail.com>]

Christopher Li pisze:
> 
> EXPR_BINOP used in normal operation require two operands.
> e.g.  a + b, a | b
> Both operands will get evaluated.
> 
> EXPR_LOGICAL using for condition branching and it has the short curcit
> behavior in mind. e.g. a || b

I found the offending block of code.  It is in expand.c, in function 
expand_logical:

       /*
        * If the right side is safe and cheaper than a branch,
        * just avoid the branch and turn it into a regular binop
        * style SAFELOGICAL.
        */
       if (rcost < BRANCH_COST) {
               expr->type = EXPR_BINOP;
               rcost -= BRANCH_COST - 1;
       }

After removing these lines, everything works fine.

But I guess that there must have been a reason to add them in the first 
place.  I see it checking the cost of the operation, but I don't know 
why somebody assumes that it would be safe not to make a branch. Does 
anybody know how to fix it without simply removing these lines?

Thanks!
Jacek