Hi again,
most probably this patch won't be applied either, nevertheless, I
inform the
problems and my own modifications....
Included in the patch (just touches sock.c and sock.h):
- Due to the semantics of the close() function, the server hangs on close()
if a client just dies (the machine just "dissapear" from the network)
and
there is still data to send in the TCP buffers. This is avoided specifying
"NO_LINGER" in the socket, which means that the socket will be closed
in
background.
- Added KEEPALIVE to avoid sucky clients.
I've checked icecast1 and squid (both are rock solid) and they include those
two options. I replicated them...
Regards,
--
ricardo
"I just stopped using Windows and now you tell me to use Mirrors?"
- said Aunt Tillie, just before downloading 2.5.3 kernel.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: patch3.diff
Type: text/x-diff
Size: 1454 bytes
Desc: patch3.diff
Url :
http://lists.xiph.org/pipermail/icecast-dev/attachments/20020301/2ccb1c5b/patch3.bin
> most probably this patch won't be applied either, nevertheless, I inform the > problems and my own modifications....Why wouldn't we apply this patch? This one is mostly obvious, and we thank you for catching it. Some of your other patches change quite big things. If I accepted every patch that was sent to me without some research and effort, icecast would fall into ruin very quickly. This is not to say that your patches are bad, but that no one is as familiar with the goals and implementation of icecast as I am. I want to make sure to preserve those goals and keep the implementation consistent and clean. Sometimes this means that patches take longer to process. Patches like this one which fix fairly obvious bugs are easy to apply and simple to digest. Patches that add large features like static streaming (one of the patches in my queue for a while) and source buffering take me some time. Patches that fix issues that I haven't experienced and may have side effects also take some time. Not everyone here cares about Windows compatability, and so often it is me that does the testing to make sure patches sent in work fine for the other platforms too. Please don't be discouraged. I'm busy and slow on this front, but I will try to get better and faster. I appreciate you sending patches to the list where they can get some feedback and testing regardless of whether I've looked that them or commited them yet. jack. --- >8 ---- List archives: http://www.xiph.org/archives/ icecast project homepage: http://www.icecast.org/ To unsubscribe from this list, send a message to 'icecast-dev-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
> - Due to the semantics of the close() function, the server hangs on close() > if a client just dies (the machine just "dissapear" from the network) and > there is still data to send in the TCP buffers. This is avoided specifying > "NO_LINGER" in the socket, which means that the socket will be closed in > background. > > - Added KEEPALIVE to avoid sucky clients.Done and committed. jack. --- >8 ---- List archives: http://www.xiph.org/archives/ icecast project homepage: http://www.icecast.org/ To unsubscribe from this list, send a message to 'icecast-dev-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
On 01/03/02 00:44, Jack Moffitt shaped the electrons to say:> > most probably this patch won't be applied either, nevertheless, I inform > > the problems and my own modifications.... > > Why wouldn't we apply this patch?It was just an encouraging joke to myself :-) -- ricardo "I just stopped using Windows and now you tell me to use Mirrors?" - said Aunt Tillie, just before downloading 2.5.3 kernel. --- >8 ---- List archives: http://www.xiph.org/archives/ icecast project homepage: http://www.icecast.org/ To unsubscribe from this list, send a message to 'icecast-dev-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.