On 30.08.2007 03:54, matt wrote:
i have noticed several patches signed off on the list that change data types from uint16_t to u16 and the like (8 .. 32 bit signed and unsigned) and was wondering why ? i thought uint16_t was c99 standard ? i asked the same thing on sdl's mailing list and was told that not all compilers support it. with sdl that seems right since its a platform independent lib. with linuxbios it seems that is not the case, i assume most or all would be using gcc.
There are several reasons: * The Linux kernel uses u8, u16 etc. * u_int32_t and uint32_t are both allowed in C99 (I have not verified that) and we'd have to choose * I can't remember whether it's u_int32_t, uint32_t, u_int32t etc. * The _t at the end is superfluous and ugly * The underscore disease is hampering all of them * Calling unsigned char an int just because it sounds cool is stupid * Unneeded redundancy * u8, u16 etc. are shorter * Less typing needed for u8 etc. * Screen estate requirements are less for u8 etc. * uint8_t looks a lot like unsigned_int_well_not_really_it_s_char_that_s_why_we_included_the_8 * LinuxBIOS uses u8 in the BIOS codebase which is neither userspace code nor can it use libraries.
only reason i ask is that a standard header was added years ago and i still see custom data types. just trying to understand that better.
I hope my opinion piece above provides sufficient reasons. ;-)
To comfort you, such switchovers are not planned for the userspace utilities we ship. Yet.
Regards, Carl-Daniel