Hi Dustin, I checked the thread you pointed out. As you can see, my patch is simpler than one on that thread and it wouldn't uglify the code so much. However, if memcached project decided to stick with C99 and drop C90 support, okay, fair enough. I know I'm a minority still using a C90 compiler. In that case, I'd suggest dropping C90 support completely. The code like below made me think it still supports old compilers...
/* FreeBSD 4.x doesn't have IOV_MAX exposed. */ #ifndef IOV_MAX #if defined(__FreeBSD__) || defined(__APPLE__) # define IOV_MAX 1024 #endif #endif /* If supported, give compiler hints for branch prediction. */ #if !defined(__GNUC__) || (__GNUC__ == 2 && __GNUC_MINOR__ < 96) #define __builtin_expect(x, expected_value) (x) #endif Cheers, Akira On Sun, Sep 28, 2008 at 3:10 AM, Dustin <[EMAIL PROTECTED]> wrote: > > > There was a similar patch that came through recently. I believe the > consensus was to not put a lot of effort into supporting older > systems: > > http://groups.google.com/group/memcached/browse_thread/thread/a546cee63d0afc9f/4e15305e4db1de48 > > On Sep 27, 5:54 am, "Akira Kitada" <[EMAIL PROTECTED]> wrote: >> Hi, >> >> The file attached fixes compile errors and warnings that appear when >> building with a C89 (ISO C90) compiler. >> >> - moved variable declarations to the start of a block >> - fixed malloc.h warning on old FreeBSD >> - changed to use inttypes.h if it is available and stdint.h is not. >> - changed to use zero-length arrays when using old gcc that does not >> support a flexible array member yet. >> http://gcc.gnu.org/onlinedocs/gcc-4.2.4/gcc/Zero-Length.html >> >> I'm not sure whether C89 compiler is still supported by memcached or >> not, but I think >> it's good thing to support broader range of compilers as long as the >> changes will not damage the readability much. >> >> Any comments? >> >> iso_c90_compat.diff >> 4KViewDownload >
