Andre, honestly, I have been unable to reproduce the bug, but there is a point in what you said regarding the CCID-split patch. In fact, you identified a real bug. I have looked at your application: * the server listens, accepts and * then writes to the client This is bi-directional mode. In TCP no problem. In DCCP that patch has disabled bidirectional mode since at that time there were strange interactions. In short: the only mode that is currently supported is * server listens, accepts * and only reads o client connects and o only writes That works with the test applications (ttcp and iperf) since this is exactly the way they operate. But it constrains socket programming. I will ask Arnaldo to revert all these stupid changes that I caused. This will allow full bidirectional connections in the first place. I have looked at the CCID3 code, it seems to be ok with that. Can you confirm that, when the `culprit' patch is reverted (as per the first email), the bug disappears? That would double the reasons. In any way, thanks a lot for identifying this obvious restriction, this has been a great help. I think it is great that real applications appear and I would like to use this paraslash application. If there are any other problems, please do email to this list. Thanks again Gerrit - To unsubscribe from this list: send the line "unsubscribe dccp" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
