it's getting there. the tarball is up on the development server (its a ppro 233 so be patient). I update it constantly.
the dedicated server for linux does compile fine and runs. its tough to troubleshoot however, hence the work with the bots.
the whole network thing is done, just in the process of finding errors in the netcode. what the q3 engine does in 2.5k/sec per player i do in 1/10th that, thanks to vic for the huffman encoding on the packet data. unfortunately for me, the network infrastructure is delicate and i do use it to its maximum. right now the worst that can be said is during a melee you will see msg's about unreliable packets being dropped, which I can surpress because it doesn't make a difference for prediction until a reliable packet is dropped.
but...
its ready, but like i posted below I am still finding bugs. the AI works fine, they don't grapple that great yet but they do go through all the weapons/classes so it makes for some fun time to spend when bored, especially with a bitch gunner that switches weapons depending on stuff. Each client carries around an ent->ai.enemy array description in them, so the bots will learn and share who does what, friend and foe. if you are a grapple monkey they are gonna act different then if you aren't, etc. always adapting. its a hack but it keeps me interested while testing for bugs and since it's not creating new ones, it works out ok. i figured out how to cheat in c, took 3 days to come up with this one function statement that dwarfs my mind. it's in the topic of #shadowspawn, its a hack and 1/2 using memory management to copy a subset array over to a fake ent, since in memory the calling function is only using that array i'm copying over. fugly to some, overlycomplicated for others, but works so its ok by me.
dunno what else. just did a client-side entity physics system, collision detection isn't perfect _yet_ for the debris but it's just a matter of finding a decent usable fraction (sometimes debris will fall through floors). stuff like doubles might need to be used in some calculations but people have mixed feelings.
still cleaning up code, making netcode more efficient, and then finding a RH box because debian doesn't like dynamic compiles and has to be static; tho since i did clean it all up for static and it's mundo efficient, i'd still like final on a RH box.
