On Mon, Nov 3, 2014 at 3:41 PM, Branko Čibej <br...@wandisco.com> wrote:
> On 28.09.2014 19:56, stef...@apache.org wrote: > > Author: stefan2 > > Date: Sun Sep 28 17:56:01 2014 > > New Revision: 1628093 > > > > URL: http://svn.apache.org/r1628093 > > Log: > > Support FSFS format 7 commits in load balanced mixed-architecture > clusters. > > > +/* Write VALUE to the PROTO_INDEX file, using SCRATCH_POOL for temporary > > + * allocations. > > + * > > + * The point of this function is to ensure an architecture-independent > > + * proto-index file format. All data is written as unsigned 64 bits > ints > > + * in little endian byte order. 64 bits is the largest portable integer > > + * we have and unsigned values have well-defined conversions in C. > > + */ > > +static svn_error_t * > > +write_uint64_to_proto_index(apr_file_t *proto_index, > > + apr_uint64_t value, > > + apr_pool_t *scratch_pool) > > +{ > > + apr_byte_t buffer[sizeof(value)]; > > + int i; > > + apr_size_t written; > > + > > + /* Split VALUE into 8 bytes using LE ordering. */ > > + for (i = 0; i < sizeof(buffer); ++i) > > + { > > + /* Unsigned conversions are well-defined ... */ > > + buffer[i] = (apr_byte_t)value; > > + value >>= CHAR_BIT; > > Does APR guarantee that sizeof(apr_byte_t) == CHAR_BIT? > No. It guarantees that sizeof(apr_byte_t) == 1 (you probably meant to say that). APR.H on all platform defines it as 'unsigned char' - as opposed to apr_uint16_t and friends, which could be anything. -- Stefan^2.
