This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Assertify rs6000


On Fri, Apr 22, 2005 at 03:28:20PM +0100, Nathan Sidwell wrote:
> David Edelsohn wrote:
> >	This generally looks okay, but I found some changes that look like
> >typos.
> >
> >	The free-standing block in num_insns_constant does not look correct.
> >
> >!       case CONST_DOUBLE:
> >! 	if (mode == SFmode)
> >! 	  {
> 	    //first bit & return
> >! 	  }
> >  
> >  	{
> 	  //second bit
> 
> I know that bit's correct -- I checked it several times.  The original
> code looked like,
> 	if (code == CONST_INT)
> 	  something
> 	else if (code == CONST_DOUBLE && mode == SFmode)
> 	  first bit & return
> 	else if (code == CONST_DOUBLE)
> 	  second bit
> 	else
> 	  abort
> 
> Admittedly the freestanding block for 'second bit' _looks_ odd.  To
> remove that I'd have to move the variables declared within it up to an
> outer scope.  I can do that if you want.

I generally keep the else in this case.  Yes, it doesn't really
add anything to the control flow, but it makes the brain hurt less.


r~


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]