RE: Tests 19 and 704 unreliable on FreeBSD after 02fbc26d5 (connect: Add connection delay to Happy Eyeballs)
Date: Sat, 16 Nov 2013 23:05:12 +0000
On Sat, 16 Nov 2013, Fabian Keil wrote:
> > Cool - I think it would be best to fix the issue in singleipconnect()
> > where it returns CURLE_COULDNT_CONNECT as I think it will make
> > the code more readable and means that sockp doesn't need to be
> > unnecessarily set and returned.
> > If I put together a quick patch - would you be able to test it for me
> > please?
> I tested the attached patch which should do what you describe and it
> works as expected. If I misunderstood, I can also test your version.
Thank you - patch enclosed.
I didn't mean any disrespect and my apologies if I have misunderstood where
the problem is.
My point really was that if singleipconnect() is failing (and I'm not too
sure why connect() would fail with an result code of -1 and leave the socket
open - except in the switch statement in that function) then I feel that
singleipconnect() should close the socket rather than have one of the
calling functions do the work.
Received on 2013-11-17