netutils use of socket descriptor

[Hiren Patel]

Thomas Guyot-Sionnest wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Ethan Galstad wrote:
>> Thomas Guyot-Sionnest wrote:
>>> On 15/06/09 02:27 PM, Thomas Guyot-Sionnest wrote:
>>>> A better way of doing this is to send a connect for the next ip in list
>>>> every n miliseconds (where n is a small fraction of the timeout, you
>>>> could for example take half of the timeout and divide it by the number
>>>> of IP), use the first connected socket and clean up the rest.
>>>
>>> Just for clarification I'm not necessarily saying that we should do it
>>> this way, but that if we really want to try all IPs that's the way
>>> (IMHO) it should be done.
>>>
>>> The attached patch removes the useless loop. There's no real-world
>>> testing besides the compilation/tests.
>>>
>> Thanks - patch is in CVS.
> 
> While the above patch doesn't change any behavior, I was wondering if it
> could loop in case we get a mix of IPv4/v6 addresses and one of the two
> isn't supported. It looks like it would return EINVAL (Unknown protocol,
> or protocol family not available.) in that case, but I'm not experienced
> at all in sockets...
> 
how would you suggest we could test this?
if you have a method for checking this behavior, I don't mind simulating 
it and seeing what the behavior is. I have not done socket related stuff 
either.

------------------------------------------------------------------------------
Are you an open source citizen? Join us for the Open Source Bridge conference!
Portland, OR, June 17-19. Two days of sessions, one day of unconference: $250.
Need another reason to go? 24-hour hacker lounge. Register today!
http://ad.doubleclick.net/clk;215844324;13503038;v?http://opensourcebridge.org