On Wed, 30 May 2001, Sterling Hughes wrote:

> > I can see that there's gonna be an extra copy operation that could be avoided
> > if we offer a different set of arguments to the callback function (which
> > indeed can be discussed).
> >
> > Is that what we should discuss?
>
> yep, it would be nice if this could be changed...

[re-iterating the subject a little]

In cases where the application will still have the contents of the file even
efter the callback function is returned, there is no point in copying the
data to libcurl's internal buffer only to write it to the network layer the
next moment.

We could still offer the current read callback for old times sake and for all
those cases where the above statement is not true.

It could possibly look like this:

   size_t read_file(void *clientp, size_t size, unsigned char **bufptr);

An application would then:

 1 - use the clientp for whatever purpose it feels like

 2 - set the *bufptr to point to the start of the buffer piece to upload next

 3 - return 'size' or less bytes (the reason for this maximum amount is
     merely for blocking purposes, so that we don't write to write X*1000
     bytes to a socket when we only do X*10 bytes per second, as that would
     make a very long block with bad a lousy progress meter as result among
     other things)

Thoughts?

-- 
     Daniel Stenberg -- curl dude -- http://curl.haxx.se/
_______________________________________________
Curl-library mailing list
Curl-library_at_lists.sourceforge.net
http://lists.sourceforge.net/lists/listinfo/curl-library