I really don't understand what we meant to accomplish on Unix, it was before I was proficient on that platform.
AIUI it meant apr_off_t was large enough to hold my file pointer, which was how it was implemented on Win32. We explicitly used a 64 bit type to ensure files were addressable. As I've come to understand it, some *nix implementations took this as the presence or absence of a specific toggle availability, which is incorrect. At this point, I'm unsure how it is used in practice. I think we need more input from users and implementors in this discussion. Such public-facing flags were meant to describe apr's behavior, not the underlying OS sillyness. On Mar 18, 2017 6:47 PM, "Nick Kew" <n...@apache.org> wrote: > Running make test on 64-bit Linux, I got an unexpected: > testlfs : |Line 345: Large Files not supported > SUCCESS > > Investigating config.log, I see the test program failed on: > | if (sizeof(off64_t) != 8 || sizeof(off_t) != 4) > | exit(1); > > OK, sizeof(off_t) on 64-bit is 8, so that line fails. > > Presumably this is just a slightly misleading message > from the test suite, and LFS is a no-op. Or am I > missing something? > > -- > Nick Kew >
