[ENet-discuss] Problem with fragmentation and reassembly of bigenet packets

[RAULO-KUMAGAI Emmanuel IT&L@bs]

Thanks for your replies!

Ruud van Gaal a écrit :
> Especially with packets >2Kb, as UDP only allows packets upto around 
> 1,5Kb. After that, Enet starts splitting them up and rejoining them at 
> the client (receiving) end, so it's important to keep calling 
> enet_host_service().
I also thought at first that the MTU could be the issue but 2KB packets 
work OK, which made me think the problem was not there.

>     *Van:* enet-discuss-bounces at cubik.org
>     [mailto:enet-discuss-bounces at cubik.org] *Namens *Nuno Silva
>     **
>     Even if Client-Only you should call enet_host_service, since it
>     takes care of transferring packets back and forth. Check if that
>     helps.
>
Now if enet_host_service() must be called on the sending side, this 
means I better produce data and send it in separate threads, right?
Do you have suggestions on how to minimize the latency of packets then?
I mean, if the thread actually sending data is inside 
enet_host_service(), how do I get it to return from that function and 
send the next packet as soon as it is available?

Thank you for your support,

--
Emmanuel


-------------- next part --------------
A non-text attachment was scrubbed...
Name: emmanuel_raulo.vcf
Type: text/x-vcard
Size: 363 bytes
Desc: not available
URL: <http://lists.cubik.org/pipermail/enet-discuss/attachments/20100429/961aa657/attachment.vcf>