On Tue, 30 Oct 2007, Georg Lippitsch wrote:

> My suggestion is: It should be possible to return 0 from readfunction, but
> keeping connection alive (NOT cause it to stop the current transfer).
> Instead, there should be a spezial return code that signals end-of-file.
>
> Of course, this might bring up some problems with backward compatibility.
> So, how about adding a new option that tells curl whether to break on
> 0-return or not?

Why 0? Why not just a CURL_MAGIC_READ_IS_NOW_PAUSED define, that certainly
won't cause any backwards compatibility issues?

-- 
  Commercial curl and libcurl Technical Support: http://haxx.se/curl.html