[ENet-discuss] ENet 1.0 "preview"

[Lee Salzman]

So, I finally got off my butt and committed all the fixes that have been
lying around my hard drive into CVS. I will disclaim, however, that I
have not tested these fixes yet, and all I know is that they compile,
but should otherwise be innocuous enough to not cause any huge bugs,
althought I am sure there is probably atleast one waiting to strike.

New stuff that is in here:

1) enet_peer_disconnect_now()
    This sends an unreliable disconnection request and immediately
tears down the connection. No disconnect event is generated on the
calling side, the the receiving end is aware of this. In the worst case,
if the disconnect packet never gets there the connection will time out
eventually. This should be useful for many situations.

2) ENET_PACKET_FLAG_UNSEQUENCED
    This flag marks a packet as unsequenced, that is, ENet is free to
deliver the packet in any order it desires, and this so this results in
lower packet receipt latencies where sequencing is unnecessary. This
only works on packets that are not marked as reliable, since the
reliability mechanism depends on sequencing to work.

3) Brad Monahan's timer resolution fixes for Win32

4) enet_initialize_with_callbacks()
    This allows you to pass in callbacks to be used in place of
malloc(), free(), and rand() for those who wanted to fill those pieces
in themselves.

5) There are probably some other miscellaneous fixes in there that i've
long since forgotten I added. Consider them easter eggs. :)

What ISN'T there:

1) Documentation for the new stuff.
2) Bug testing.
3) Sane timeout limit handling, as hinted at in another thread.

If anyone would like to help with these, that'd be cool. :)

When those 3 things are fixed, I'd basically declare this 1.0 and
release it as such.

My excuse: it's my last semester of university, and I'm horribly busy
trying to pass classes and graduate all the while writing my thesis.
Time for ENet has been somewhat lacking.

Lee