In the 10 minutes or so I spent poking through the code (deadlines at
work don't allow to me poke longer, since I have to move on to other
things) it looked like there was also some connection verification code
that doesn't get run for multi-handle connections. The multi-handle
ones were the *only* ones that I had problems with prior to patching the
source myself. I would've liked to investigate that a bit further, but
it would have to be on my own time. And that's something I won't have a
lot of for at least a week or two.

- jk

Daniel Stenberg wrote:

> On Wed, 9 Nov 2005, Joe Khoobyar wrote:
>
>> Any thoughts on how I can get this to work? The original poster's
>> code for fixing this bug wasn't included in the message below and I
>> couldn't find it in the archives, so maybe there's something I'm
>> missing?
>
>
> Shmulik Regev provided his version of a similar work-around as you did
> in this mail:
>
> http://curl.haxx.se/mail/lib-2005-11/0056.html
>
> Still: why does it help to wait a very short while? Can't we instead
> have the code retry the operation in case it gets that failure instead
> of unconditionally always sleep a little while?
>
Received on 2005-11-10