[U-Boot] [Patch v3] Consolidate bool type

[Allen Martin]

On Thu, Jan 24, 2013 at 10:14:37AM -0800, Scott Wood wrote:
> On 01/24/2013 12:03:49 PM, York Sun wrote:
> > On 01/24/2013 09:54 AM, York Sun wrote:
> > > On 01/24/2013 09:34 AM, Allen Martin wrote:
> > >> On Wed, Jan 23, 2013 at 02:05:26PM -0800, York Sun wrote:
> > >>> On 01/23/2013 02:02 PM, Scott Wood wrote:
> > >>>> On 01/23/2013 04:01:49 PM, York Sun wrote:
> > >>>>> On 01/23/2013 01:52 PM, Scott Wood wrote:
> > >>>>>> On 01/23/2013 03:46:04 PM, York Sun wrote:
> > >>>>>>> On 01/23/2013 01:41 PM, York Sun wrote:
> > >>>>>>> I should put RFC in the subject as I am not able to compile  
> > all ARCH
> > >>>>>>> myself.
> > >>>>>>
> > >>>>>> So how do you see this patch becoming non-RFC?  I think most  
> > people
> > >>>>>> don't have every single toolchain.  You should at least get a  
> > toolchain
> > >>>>>> for a couple major architectures such as ARM.  Usually RFC is  
> > for when
> > >>>>>> you know the patch has issues, and don't want it applied yet  
> > even if
> > >>>>>> nobody else finds fault with it.
> > >>>>>>
> > >>>>>
> > >>>>> I know this version has problem. I am hoping more people get  
> > involved
> > >>>>> and test what they can. I tried the arm toolchain you pointed  
> > to me. I
> > >>>>> couldn't run MAKEALL for arm.
> > >>>>
> > >>>> Even before your patch, using USE_PRIVATE_LIBGCC?  How many  
> > boards failed?
> > >>>>
> > >>> I don't know. Wolfgang brought it. Allen kindly offered help to  
> > verify
> > >>> on tegra which uses USE_PRIVATE_LIBGCC.
> > >>>
> > >>
> > >> Built and tested on tegra, no problems.  I'm still seeing a lot of
> > >> references to  TRUE/FALSE even after this patch though:
> > >>
> > >> $ git grep -e FALSE -e TRUE | awk 'BEGIN {FS = ":"} {print $1}' |  
> > sort | uniq
> > >>
> > > Thanks, Allen. A long way to go.
> > >
> > 
> > I thought I have replaced all #define, enum, typedef. I have left  
> > alone
> > those FALSE, False, false but add define like this
> > 
> > +#include <stdbool.h>
> > +#define TRUE true
> > +#define FALSE false
> > +#define True true
> > +#define False false
> > 
> > Isn't that enough?
> 
> It's enough to make it build, but it would be better to fix the users.
> 

For code consistency I think it would be better to force everything to
use the lower case "true/false".  Otherwise this patch does nothing to
force new code to use the standardized version.

There may be some exception we want to make for unmodified 3rd party
libraries that are checked in, like lib/bzlib which uses True/False,
otherwise it makes it hard to take new code drops of those things.

-Allen
-- 
nvpublic