On Mon, 28 Mar 2005, Augustus Saunders wrote:

>> Then in what way doesn't FAILONERROR do what you want? And if not, isn't
>> returning an error from a header callback good enough to add a "custom"
>> abort?
>
> I can't quite tell what FAILONERROR is doing. What I want is for the upload
> to abort if an error is returned. Especially since an "error" will often be
> 407--I can't make my users wait for 3 minutes of uploading only to pop up a
> proxy auth dialog and then try again! No point in transmitting 50 megabytes
> if the server ain't taking it. And yes, I can do some magic in the header
> callback, but this seems so obvious I kind of assumed there was already an
> option for it.

This is where "Expect: 100-continue" should be used I guess. libcurl doesn't
add that by default for plain posts. Otherwise there's no way for the server
to signal error until you've sent the whole post...

> In response to your second email, yes, I can try out the nightly build
> tomorrow. But I'm prepping a release of my software, so I'll need to know
> exactly what fixed the problems and make those changes against 13.1--I can't
> risk any other changes.

You can always see the latest CVS changes here:
http://curl.haxx.se/auto/cvshistory.html

And from there, you can easily browse diffs of the changes done to single
files. In this case, the changes done to HTTP matches what you explained
previously: http://cool.haxx.se/cvs.cgi/curl/lib/http.c.diff?r1=1.265&r2=1.266

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