On Thu, Mar 4, 2010 at 5:24 AM, Vincent Torri <vto...@univ-evry.fr> wrote: > On Fri, 26 Feb 2010, Gustavo Sverzut Barbieri wrote: > >> On Fri, Feb 26, 2010 at 8:12 AM, Vincent Torri <vto...@univ-evry.fr> >> wrote: >>> >>> On Fri, 26 Feb 2010, Gustavo Sverzut Barbieri wrote: >>> >>>> On Fri, Feb 26, 2010 at 4:22 AM, Vincent Torri <vto...@univ-evry.fr> >>>> wrote: >>>>> >>>>> On Thu, 25 Feb 2010, Enlightenment SVN wrote: >>>>>> >>>>>> Log: >>>>>> TRUE/FALSE are gone, use EINA_TRUE/EINA_FALSE instead. >>>>>> >>>>>> Author: barbieri >>>>>> Date: 2010-02-25 16:59:11 -0800 (Thu, 25 Feb 2010) >>>>>> New Revision: 46500 >>>>>> >>>>>> Modified: >>>>>> trunk/eina/src/include/eina_types.h >>>>>> >>>>>> Modified: trunk/eina/src/include/eina_types.h >>>>>> =================================================================== >>>>>> --- trunk/eina/src/include/eina_types.h 2010-02-26 00:57:20 UTC >>>>>> (rev 46499) >>>>>> +++ trunk/eina/src/include/eina_types.h 2010-02-26 00:59:11 UTC >>>>>> (rev 46500) >>>>>> @@ -214,25 +214,13 @@ >>>>>> #endif /* ! __GNUC__ && ! _WIN32 && ! __SUNPRO_C */ >>>>>> >>>>>> >>>>>> -/* remove this TRUE/FALSE redifinitions */ >>>>>> - >>>>>> /** >>>>>> - * @deprecated Use #EINA_TRUE instead. >>>>>> - */ >>>>>> -#ifndef TRUE >>>>>> -# define TRUE 1 >>>>>> -#endif >>>>>> - >>>>>> -/** >>>>>> - * @deprecated Use #EINA_FALSE instead. >>>>>> - */ >>>>>> -#ifndef FALSE >>>>>> -# define FALSE 0 >>>>>> -#endif >>>>>> - >>>>>> -/** >>>>>> * @typedef Eina_Bool >>>>>> * Type to mimic a boolean. >>>>>> + * >>>>>> + * @note it differs from stdbool.h as this is defined as an unsigned >>>>>> + * char to make it usable by bitfields (Eina_Bool name:1) and >>>>>> + * also take as few bytes as possible. >>>>>> */ >>>>>> typedef unsigned char Eina_Bool; >>>>> >>>>> char bitfields are an extension of gcc (and of some other compilers). >>>>> The C standard says that "fields are packed into 'storage units', which >>>>> are typically machine words". Compile a small example with gcc >>>>> -pedantic >>>>> and you'll see a warning. C99 allows _Bool, though. >>>>> >>>>> I'm not sure at all that they will take as few bytes as possible in the >>>>> sense that you implicitely say (one char -> only 1 byte in memory). >>>>> They >>>>> are always aligned to 32 bits. They just need to be consecutive to take >>>>> the less space in memory as possible. If you use some gcc options to >>>>> pack >>>>> bitfields, be aware that the library might not be working at all with >>>>> vc++ >>>>> as the way vc++ pack bitfields is different. That said, you'll >>>>> certainly >>>>> add that Windows is my territory, not yours :p >>>> >>>> damn, if it breaks then we need to convert from char to int ASAP, so >>>> we don't break the ABI after release. >>>> >>>> I'm quite sure this is in lots, lots of code paths in EFL/E17 now, so >>>> it should "just work", no? >>> >>> I've found that option to gcc (mingw): >>> >>> -mms-bitfield >>> >>> it should solve the problem when compiling the efl on Windows. >>> >>> anyway, i think that your note is not entirely correct >> >> but should we keep using it as uchar or move to uint? > > uint it the standard. Using uchar should not be a problem on Windows if I > pass -mms-bitfield. > > I'm really wondering if there is some size improvement when using a char and > not an int.
for bitfields no, but when not using bitfields, it can save 3 bytes per entry. -- Gustavo Sverzut Barbieri http://profusion.mobi embedded systems -------------------------------------- MSN: barbi...@gmail.com Skype: gsbarbieri Mobile: +55 (19) 9225-2202 ------------------------------------------------------------------------------ Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev _______________________________________________ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
