scary_pigeon

this nights test was to see if a fix to the problem we were having of bullets not appearing on each others comps worked - it did. it was related to this: the object spawns in by client instruction, then shortly after, the client arms the plane.. ..between that time, my client running on same machine as server was quick enougth to grab that planes data before it was armed - before it had completely spawned in.

hehe well how about this? slightly biased to one particular game though.. i quite like digital-overloads one.

my impression is that those cockpits are going well but probably are not lua scripted yet? I am uncertain sometimes of exact state of the project in areas that im not working on. hopefully dante will pop in and report on that. *sigh* if we had a studio i'd be able to go over and say, "what u up to?" - as for myself, The netcode continues to be buggy. As each bug is fixed - less obvious ones emerge. we seem to have cracked the bug relating to the server nolonger receiving updates. now what has become obvious is this odd bug: I log in, then dante logs in and I cant see his bullets, but he can see mine. Some bugs are quite hard to track down when they are not easily reproducable as that fault doesnt occur when two clients are run on my PC.

great to see flightsimers interested in our project :) as fates say - not yet available. soon the demo.

still bugs to tease out network wise. its really very tedius. i've spent the last 6 hours running it, waiting for it to fall over at some random point - try figure out where about in the code this happened then move try something to maybe fix it. right now, i am again waiting to see if something goes wrong. for the last 15 minutes ive had two clients logged in and all seems to be well. haha. as i sat there waiting for odd things to happen. i noticed one of the planes vanished from view. This is odd. hopefully tomorrow I will fix it. It is important that these netcode bugs be fixed early because the netcode approach is the heart of our system for both multiplayer at single player AI driven ones.

looks splendid.

bad, news, that didnt work as planned. until now i'd only tested the code on my home pc and sometimes you get issues that low latency and very high bandwidth can hide. So for hours I expect now to be sulking infront of my pc trying to find the niggly little bugs. edit: and the hours pay off, we can see each other - it works quite well - im in the UK, dante is in brazil.

the last few week or so has been concentrated code wise on server-client event stuff. it progresses quite well, we are on the verge of at least the developers, being able to shoot at each other across the net. i'm waiting for a fairly rare period when dante is actually at home rather than is main job, so he can try the client to fully check the multiplayer stuff. Its just too awkward to test on one machine.

http://www.gorlice.net.pl/~rybak/luaide/ luaide is a worthwhile thing I found.

the english is fine. appart from waypoints being just positioning data, attached is also an arbitary name. the script can then check the word against some behaviour triggers. what you say is possible, it already works in this way.

movie: click picture above some readers of this thread will have seen such movies before of me chasing another aircraft, however, now the code is significantly different. there is a server client approach. on this server, bots or other human players can do their thing and this movie is where I am testing this out - in a modified cockpit shell prog the netcode - were I follow sea harrier. where double teaming a poor but nimble skyhawk over featureless infinite grass. next steps go like this: add in the bullets - on the server I can see the bulets wizzing about - this is not yet implemented visually client side. this is next step. after that: implement collision detection and event system for this server-client approach. then the rest of it. place the landscape back in and create some more interesting missions. (the landscape is not in this movie because it is quite slow to load when I am just wanting to tweak netcode, fm, cockpit avionics)

hehe.

enginewise, theres no hard limit to texture size i'm sure when people have the game in their posession they will be doing nutcase 5000x5000 pixel textures for aircraft.

gay: its got a non blocking server with a finite state machine approach to avoid lock ups. hopefully the demo, after closed testing will wow people. a little like how IL2 demo wowed me.

an update. last few weeks we've been working towards a demo. it has been necessary to do a lot of work on the network code. in some ways the netcode is quite good. it splits the world into sectors - events and information is spread between other clients - as well as from server run air/land/sea bots - based on which objects need to know. i realised that a way to accelerate programming of all this is to continue with our savior - lua scripting. using a sort of text chat approach its possible to script program a lot of the net code - particularly the events and many of the events we havent thought up yet this system is now feeling complete - fully tested in shells. next step. apply all this in a graphical client. not long now - demo will be soon

blimey. that is scenario that has to happen. I was not really concious off it till now. I am wondering if amongst this effort are there any people who know LUA. I think LUA wasnt around when falcon4.0 came into being and it really very nice. our cockpit, sound fx, dials, electronic displays, keyboard interaction, flightmodel etc, etc - all goes through LUA - so it might be possible for JT compatable cockpits to be created.

i was trying the demo before.

yup. ASDA owned by corporate facist Walmart.

ahh brilliant, great to see you idle - what are you up to these days?

cheers, and welcome to the forum.

most of the server net code is now complete and tested. some things remain. then its case of make a graphical client - rather than the debugclient. once that is done, the demo to be released soon becomes days away. or to closed beta types anyway. :P the screenshots dante shows are WIP experiments - they lack stuff that makes it look good, proper non place holder textures, detail textures. What is technically good is the undulating terrain that match satalite height data, you dont really get any better than this. - this is different than the more usual heightmap approach were terrain heights are at a rather corse(sp?) resolution of 256 levels.

lua is awesome. using it for our project :-) it at the moment handles AI, physics, mission stuff, cockpit hud functionality. I imagine that in lomac it is similarly useful.

:)

at the moment im continuing to do the netcode. the overall approach is to kill two birds with one stone. the netcode is also the server code - the player bubble cleverness - that lets you have a multitude of objects at reasonable speed is on the server. the client has a bubble which it sees from that. before this can happen some basic netcode needs doing over and above whats been done in the project before. simple things like spawn points. about this time last year me and dante were flying over the net - a sort of learning experience that was - now we do it right and complete. the idea of doing it this way is that neither campaign nor netcode ends up as some sort of tacked on at the end half done thing that doesnt work properly.

well at least the war is an example of how wars should be fought. when these argentine imperials (im sure that term will amuse our argentine friends here) came in, they didnt go round ravaging the place. they shat in some peoples baths and stuff for not feeding them, but there was none of the usual harshness you hear of in the news. near zero civ casualties. bit of a shame for the argentine conscripts though - who should really have been in college or something. but as wars go, it was cleanly done. none of this 100,000 civ casualties that the lancet projects in iraq and all that.