On Dec 16, 2010, at 4:38 PM, William A. Rowe Jr. wrote: > On 12/16/2010 3:36 PM, Jim Jagielski wrote: >>> >>>> + # where int and long are the same size. Use the longest >>>> + # type that fits >>>> + if test "$ac_cv_sizeof_off_t" = "$ac_cv_sizeof_long_long"; then >>>> + off_t_fmt='#define APR_OFF_T_FMT APR_INT64_T_FMT' >>>> + off_t_strfn='apr_strtoi64' >>> >>> This is bad, no? We don't know that long_long and off_t aren't 128 bytes. >>> It seems better to use the explicit "ll" format here instead of the value >>> reserved for 64 bit ints. >>> >> >> All I did was re-arrange the order... > > Not arguing, suggesting that the thorough test is either to compare the > ac_cv_sizeof_off_t to 8, and then use APR_OFF_T_FMT, or failing that, instead > see if it matches long_long, and use an explicit "ll". >
Oh, yeah. Well, it's not only the format but everything as well... After all, if long long is 128bits, you don't want to use apr_strtoi64 either. I would suggest that we tackle that issue separately?
