Re: [E-devel] Killing off Eina_Bool

[Daniel K. <qua...@gm...>]

2014-08-29 1:09 GMT+01:00 David Seikel <on...@gm...>:

> On Thu, 28 Aug 2014 22:23:03 +0100 Tom Hacohen <to...@st...> wrote:
>
> > On Thu, Aug 28, 2014 at 7:51 PM, Cedric BAIL <moa...@gm...>
> > wrote:
> >
> > > On Aug 28, 2014 6:10 PM, "Daniel Kolesa" <qua...@gm...> wrote:
> > > >
> > > > 2014-08-28 17:06 GMT+01:00 Cedric BAIL <ced...@fr...>:
> > > >
> > > > > Hello,
> > > > >
> > > > > On Thu, Aug 28, 2014 at 5:12 PM, Tom Hacohen
> > > > > <tom...@sa...> wrote:
> > > > > > It's been ages since we last complained about the existence of
> > > > > > Eina_Bool. We all hate it and think it's useless. It's too
> > > > > > long to
> > > type,
> > > > > > and redundant. "bool" has been available for ages now. Same
> > > > > > goes for "true" and "false". I think we should kill Eina_Bool
> > > > > > and start using those. It doesn't have to be a sed (though we
> > > > > > could sed it out from
> > > all
> > > > > > of our sources), it can be gradual.
> > > > > >
> > > > > > If someone thinks there's a platform out there that doesn't
> > > > > > have
> > > "bool",
> > > > > > we can just add a platform check and define it if missing.
> > > > > >
> > > > > > What do you think?
> > > > >
> > > > > That it is globally a bad idea. The bool type should not be
> > > > > used in any of our public header (due to how poor the standard
> > > > > define it). It also should not be used in any of our bitfield
> > > > > as nobody knows its size. So starting to use it is sure to
> > > > > trigger a mess at some point. So if it's just an issue of
> > > > > typing, make a macro in your editor and be done with it.
> > > > >
> > > >
> > > > For API it's fine. For bitfields it sucks but we can just replace
> > > > all Eina_Bool bitfields with unsigned char.
> > >
> > > It's not. It break ABI. There's no compatibility between compiler
> > > on its size. No ffi also. And yes, it would break bitfield.
> > >
> >
> > As I was telling Quaker, it's less about getting rid of Eina_Bool,
> > and more about getting rid of EINA_TRUE and EINA_FALSE in favour of
> > their nicer counterparts. Those are the ones we type all the time,
> > not Eina_Bool itself.
>
> If it's a matter of typing, why not just create macros in Eina that wrap
> Eina_Bool - ebool, etrue, efalse?  Then we can have our cake and eat it
> to.  Otherwise as others said, it's a big API break.
>

Moving from Eina_Bool to bool is mainly an ABI break, which is worse. Tom
was mainly talking about using the true/false values from stdbool.h though,
which are defined to be the same as integer constants 1/0. Thus, it's
perfectly fine to kill off EINA_TRUE and EINA_FALSE and use true/false
instead.


>
> --
> A big old stinking pile of genius that no one wants
> coz there are too many silver coated monkeys in the world.
>
>
> ------------------------------------------------------------------------------
> Slashdot TV.
> Video for Nerds.  Stuff that matters.
> http://tv.slashdot.org/
> _______________________________________________
> enlightenment-devel mailing list
> enl...@li...
> https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
>
>