On Fri, 5 Sep 2008, Dmitriy Sergeyev wrote:

>> I agree. Any chance you can write up a patch for this?
>
> ok, here is my attempt to do this. Not sure how much it is proper, at least
> it works for me. :)

There are two outstanding problems with this patch:

A) It assumes that the multi perform function is called again before the
    returned failure is noticed. This may be true in most cases, but I bet we
    can think of scenarios where it won't be called, or at least won't be
    called again for quite some time and that may end up rather strange
    behavior to applications.

B) The easy interface also needs to discover the situation and your patch
    doesn't adress that.

While B is easily solved, I'm not sure how we can fix A in a non-hackish
way...

-- 
  / daniel.haxx.se