[Opentnl-general] TNLtest is bugged

[matt <ope...@ho...>]

Hi,

There are a couple issues with the 1.4.0RC4 build for windows.  I'm
running XP and using the .NET compiler (2002, version 7.0.9466).

The problems I'm seeing are from comparing the current version of
TNLtest to the version in release 1.1.2.

1) The buildings are not sent to the client machines.  When the
buildings are generated there is no way to mark them as always in local
scope, like in version 1.1.2; and they never get ghosted on the clients.
The clients are always running around without knowledge of the building
existing.
2) The client times out after 50 seconds of inactivity.  Both the host
and client are sending activity pings to each other, but the ping
packets are getting thrown out in the "NetConnection::readPacketHeader"
function.  It happens at "pstream->decryptAndCheckHash", the packet is
too short and gets rejected by this function.  Later in the
'readPacketHeader' function though, there is a check if the packet type
is a ping packet (line 464) and it calls sendAckPacket in response.
This will never get called for a ping packet.
3) The client only receives updates of the AI positions when the client
player is moving.  As soon as the player stops, the AI updates stop as
well.  And the client's AI players remain wherever their last moved
stopped them.  It seemed that the server's client ghost was detecting
the AI players in its scope, but not sending those AI movements to the
client.

And lastly, there is a bug in the VC compiler that prevents Zap from
building.  The 'ProjectileInfo' structure contains the 'Color' struct
witch is non-aggregate, and the compiler freaks out.  The projectile
structure is aggregate, so there should really be no issues (MS bug),
but it still doesn't compile.

Are these bugs currently known or fixed?  If so can you tell me how to
get these items resolved?  I'm working on a project with similar
functionality to TNLtest and would appreciate it.

Regards,
Matt