Fragmentation (reliable or unreliable) always waits for all fragments of a packet to be reassembled before it can deliver a packet. With unreliable fragmentation, if a later (higher sequence number) packet is ready to be delivered, then any incomplete packets (missing fragments) with lower sequence number simply get dropped and not delivered. With reliable fragmentation, it would have had to wait for the earlier packet to complete instead of just dropping it in order to finally deliver the later complete packet.
On 05/31/2011 09:23 PM, Chris Jurney wrote:
Just to be clear... if I set the unreliable packet fragmentation flag, and send a large unreliable packet that is broken in two, then only one of the two pieces arrives (first half or second half) before my next packet, enet will drop them on the floor rather than deliver me half the data, correct? That's what I'd expect, just wanted to make sure I understood. On Tue, May 31, 2011 at 12:56 AM, Lee Salzman <[email protected] <mailto:[email protected]>> wrote: It is possible that you will receive duplicate or out-of-order unreliable packets. I kind of never noticed the regression in Sauerbraten because it's unreliable packet usage was never actually triggered it or had noticeable effects even if it did. On 05/31/2011 05:22 PM, Ruud van Gaal wrote: It sounds important, but I'm not sure what you mean with 'regression in packet queing'. Is it important in every-day use, or is it an uncommon scenario? Ruud On Tue, May 31, 2011 at 3:30 PM, Philip Bennefall <[email protected] <mailto:[email protected]> <mailto:[email protected] <mailto:[email protected]>>> wrote: Thanks Lee, have downloaded this and will upgrade my engine. Aappreciate all the work you put into this library! Kind regards, Philip Bennefall ----- Original Message ----- From: "Lee Salzman" <[email protected] <mailto:[email protected]> <mailto:[email protected] <mailto:[email protected]>>> To: "Discussion of the ENet library" <[email protected] <mailto:[email protected]> <mailto:[email protected] <mailto:[email protected]>>> Sent: Tuesday, May 31, 2011 4:26 AM Subject: [ENet-discuss] ENet 1.3.2 and 1.2.4 released! While I was investigating how to do unreliable fragmenting, I came across a super silly regression in unreliable packet queuing. So this release is sort of a twofer, you get unreliable fragmenting option via ENET_PACKET_FLAG_UNRELIABLE_FRAGMENT and the bug fix (so upgrading is recommended). No protocol breakage was done, so it is compatible. ENet 1.3.2 download link: http://enet.bespin.org/download/enet-1.3.2.tar.gz ENet 1.2.4 download link: http://enet.bespin.org/download/enet-1.2.4.tar.gz ChangeLog: ENet 1.3.2 (May 31, 2011): * added support for unreliable packet fragmenting via the packet flag ENET_PACKET_FLAG_UNRELIABLE_FRAGMENT * fixed regression in unreliable packet queuing * added check against received port to limit some forms of IP-spoofing ENet 1.2.4 (May 31, 2011): * fixed regression in unreliable packet queuing * added check against received port to limit some forms of IP-spoofing _______________________________________________ ENet-discuss mailing list [email protected] <mailto:[email protected]> <mailto:[email protected] <mailto:[email protected]>> http://lists.cubik.org/mailman/listinfo/enet-discuss _______________________________________________ ENet-discuss mailing list [email protected] <mailto:[email protected]> <mailto:[email protected] <mailto:[email protected]>> http://lists.cubik.org/mailman/listinfo/enet-discuss _______________________________________________ ENet-discuss mailing list [email protected] <mailto:[email protected]> http://lists.cubik.org/mailman/listinfo/enet-discuss _______________________________________________ ENet-discuss mailing list [email protected] <mailto:[email protected]> http://lists.cubik.org/mailman/listinfo/enet-discuss _______________________________________________ ENet-discuss mailing list [email protected] http://lists.cubik.org/mailman/listinfo/enet-discuss
_______________________________________________ ENet-discuss mailing list [email protected] http://lists.cubik.org/mailman/listinfo/enet-discuss
