2007/11/12, Daniel Stenberg wrote:

> >> I believe the reason is that you can provide the POST data via a read
> >> callback, with no limit on size. That's why there's also the
> >> CURLOPT_POSTFIELDSIZE_LARGE option.
> >
> > And does it actually work when using a postfieldsize greater than (size_t)-1
> > on platforms where sizeof(curl_off_t) > sizeof(size_t) ?
>
> Since the returning of data is done with very many callbacks, I don't see why
> it would be any problems to do this...

This thread is getting quite theoretical ;-) That was my reason for
asking if someone has actually tested the above mentioned condition.

I'm not saying that it doesn't work. I was just asking if someone has
actually had the time to verify that it works when using a
postfieldsize greater than (size_t)-1 on a platform where
sizeof(curl_off_t) > sizeof(size_t).

Its not a matter of trying to fight against everyone else :-)

But since at several points internal buffers are used, I have the
feeling that we are (size_t) limited. I might be completely wrong and
overlooking or missing something. That was also the reason for asking
for hard facts, practical ones not theoretical ones.

I know that I'm not providing evidence that it doesn't work. But I'm
not claiming that it doesn't work. That's another reason for me
claiming that I wouldn't change that specific code.

-- 
-=[Yang]=-