2008/8/11, Daniel Stenberg wrote:

> > There is still some logic left which was based on the direct relation
> > between curl_off_t size and large file support. This logic has to be
> > removed/changed or adjusted as appropriate. In current CVS state
> > the large file support should be considered broken until this is fixed.
>
> Indeed, they used to be tightly connected but now curl_off_t is just a 64bit
> type if available, so I think we need a new configure check to test for LFS
> so that the code can correctly do the right thing (on the few places that it
> really matters).

Yep, certainly.

Do you have something clever in mind ? Or some clues that should be
taken in consideration ? Or any general approach that you can see
would be better ?

-- 
-=[Yang]=-