Hi,

As someone might have already noticed, I finally took the route of not
including the CURL_SIZEOF_OFF_T sizeof(off_t) in the public interface
for this release.

The only benefit of including it in the public interface would be, for
configure capable systems, to enforce that the same size of off_t is
used to build the library and the apps that use the library.

Since it is not completely clear that we actually want to make this
enforcement effective, it is not done for this release.

If someone provides good arguments to enforce this setting, or if
later on it is certainly verified that this should be done in this
way, it can always be done for another release.

-- 
-=[Yang]=-