Moa

svn access would be great. Once it is up I intend trying to control things via a webservice (via JAX-WS), eg. read aircraft state, and write some commands to it. Great stuff that it still works under Java 7. I hear Eclipse has a few issues (but then, it is generally less portable than Netbeans). Will also try experiment with text rendering using textures this weekend, and give you some source you can benchmark.

VNAO website has some issues at the moment. They are being worked on. Meanwhile you can get the mod from the following page: http://www.stallturn.com/wiki/en/VNAO_mods The F/A-18F model in that mod uses the MiG-29G cockpit for the following reasons: * you can do air-to-air and guided air-to-ground with it * it displays Imperial units (knots, feet) so that you can fly the same numbers as given in official US Navy documentation (without having to convert to meters, kilometers etc in your head - which a couple of the ex-Navy pilots in VNAO hated doing). The F-14D in the mod uses the F-15 cockpit, but can drop dumb bombs and can actually destroy SAMS by diving to force the SAM into the ground and moving at high speed into the no-engagement zone of the SAM (a tactic one of our guys picked up from USN F-14 pilots and confirmed by RAAF F-111 pilots). Anyway, don't forget to check out the VNAO clips on youtube (especially those from the user 'Fumbaga'). The FC2 server that allows use of the VNAO mod is called 'stallturn.com' (disclaimer: my server), and the statistics for VNAO are at http://stallturn.com/scores/ (Java browser plugin required). It is temporarily running A-10C at the moment (because the disk in my A-10C server died) but will be back running VNAO full-time this weekend. Enjoy.

Yoda, are you caching the rendered text to a (transparent) BufferedImage? You shouldn't re-render the text if it hasn't changed (although I have a feeling whether I asked you this a year ago). Trade memory for speed. The other thing you could do is pre-render all alphanumeric characters to a texture and then use texture coordinates to select the letters. You won't get proper kerning (although machine interfaces are lousy for this anyway) but it is fast. ED seems to do this by loading characters as textures (eg for aircraft numbers etc).

I run the stallturn server. I leave it without password protection so all can fly on it. Plenty of people do, but as a consequence it does attract the teamkillers (either killing friendly units or other players) and the anti-social trolls who post inflammatory stuff in the chat window. The reason most servers are locked is to improve the experience for the players that bother to join the passworded servers (for example the 74th merely requires you to connect to their teamspeak to get the password of the day; you get fewer idiots and you have better coordination, since you're on teamspeak). So, in short, one reason servers are locked is because there are significant number of trolls who delight in teamkilling and other anti-social behaviour etc. Don't blame the server owners, they're just trying to give a better experience to the (majority) decent players [clearly I am negligent in this regard with stallturn :)]. Blame the trolls instead. If you do a little work you will find that many of the passworded servers are easy to access, provided you meet some minimal conditions (eg. joining the server's associated teamspeak, or checking a forum for NOTAMS which provide the daily password). Also, more unpassworded servers are coming online as time goes on (check out the excellent 104th A-10C server which has only been out for a couple of weeks; disclaimer, I'm a 104th member, and that server has missions that suit our more realistic tastes). @paulrkiii: kick-ass! a Hawgsmoke would rock (and thanks for the excellent checklists too).

Looking good Yoda. TCP does have checksum, but this doesn't help if packets are lost completely (depends on whether you make more than one transmission or not - if you only do one transfer then it makes no difference). Should not read back from OpenGL because the transformations are done on the GPU and stored in VRAM. Reading back from VRAM is horrifically expensive. Yeah, Java is slow for trig because it is IEEE compliant (unlike the Intel x87 hardware). You can get even faster math by doing what we used to do in the old days, create lookup tables. These days the RAM use is negligible compared to textures etc. so they are a good option for extreme performance. It sounds like you have your performance good enough as it is already though.

May I make a suggestion? The last time I checked LEAVU2 (and perhaps things have changed since then) the exported data is streamed with a carriage return after each item. I suggest that each 'group' of data be exported as comma separated values with a carriage-return at the end of the group. There are two reasons for this suggestion: 1) it is easy to parse the lines by eye, as they fill a line, and 2) if one piece of data is corrupted on a line the other lines are not affected (unlike the on-per-line streaming where a single corrupted value affects all subsequent value). I implemented this in my own version of LEAVU2 (when I extended it for mechanical data flaps/hook/gear) and it works fine.

I hope they put at least as much effort into making the terrain look good as they put into modelling the toilets. I have X-Plane 9 and enjoy it, more-so than FSX+Acceleration. Still ain't as cool as the DCS modules or even FC2 for the multiplayer goodness. That said, I'll definitely be getting X-Plane 10. Looks like the community (and more importantly, modders) is shifting from the end-of-life FSX to far more open X-Plane. About time!

While it is only 2 MB/s up it is cable and seems to be able to handle a dozen folks before there are problems (provided everyone has reasonably stable connections, under 400 ms ping). Max is around 16.

Hi aaron886. I put this algorithm (and its inverse calculation) into Yoda's LEAVU2 (which he then tweaked). You can find a working (Java) implementation there. I can also supply test coordinates to check your algorithm if you like.

Great stuff redfish. There's more than a few RED units in a Soviet armored regiment, so no-one will be idle :) Kick-off is 19:00 NZDT, so probably 17:00 local if you're on the east coast of Oz (eg. Sydney).

Mission: Fulda '89 To be flown on the 'stallturn' server on Thursday (tomorrow): 2011-07-07 07:00 Zulu (19:00 NZDT) World War III has started and you're in an A-10. Fly to the Fulda Gap to defend US forces from the oncoming Soviet hordes. Full mission briefing as PDF from: http://www.mediafire.com/file/56dlen693mb1c95/StallturnMission3_Fulda89.pdf See you there!

Spasibo bolshoe Olgerd! Documentation is just what we need.

Lua and Java. There is a lot of existing structure there, so you don't have to start with a blank slate. Just extend what is already there. For Java all the tools are free (including good Integrated Development Environments like Netbeans and Eclipse). The Leavu2 source code and build project was made available by Yoda at http://www.kenai.com (http://kenai.com/projects/leavu2) but Oracle has shut that site down and moved projects to http://www.java.net (although it doesn't look like leavu2 was migrated).

Yes, those sims were great, but you also have nostalgia to blame. None of them come remotely close to the depth and breadth of A-10C. None of them required remotely the same total development time to produce (ED has been working iteratively for a *long* time to get things where they are now). This is one of the reasons why there are so few in the flight sim market.

Oh yeah, I'm sure most people have seen the community developed dynamic campaign system for IL-2 (principally built by Paul Lowengrin): http://www.lowengrin.com/news.php That campaign generator may not be perfect, but it is still vastly better than what we have now in DCS/FC2 - and doesn't require 20 hours in the mission editor to manually build a decent (non-random) mission. Can we do it (a community-build dynamic campaign system) ? Yes we can! We clearly have the talent (eg. 16th_Speed) and motivation. What we need is a stable foundation that won't change *radically* between now and the likely release date.

Just wow! That dynamic in-game engine sounds sublime. Incidentally, the way I was working was to reduce dependence on any changes the ED dev team would make. I mean, to my mind I reduced dependence on the LUA interfaces to the bare minimum and thought, "it's not like going from patch-level 1.1.0.7 to 1.1.0.8 they'd completely break the way they logged stuff, right?". Turns out I was completely wrong. With no warning the logging interface from FC2 onward has changed and the injection of customised data into the log (required since the debrief log is incomplete, without so much as a single callsign) doesn't work (for me; I can inject into the DCS.log but that log is not where I want the entries to go). How can we invest a year of our weekends and free nights into modding when at the end of it everything can change between minor patches, that don't even show up in the changelog?

Well, it's true a journey of a thousand miles starts with a single step. Turns out that the community was 1/3 of the way along the journey (post-battle stats). Although with recent changes they've been taken back to square one (unless we can find a short-cut to get us back where we were). Sure, the community is ponderous in getting places, but with EDs focus currently on patching A-10C, compatibility for Ka-50 and the possibility of a FC3 , then I'd say ED have their hands full for the medium term. The community is likely to get to the finish line first, and do a more comprehensive job (since for some people that would be their principal focus). It's very nice ED have a dynamic campaign in their plans. I for one am not going to be sitting on my hands waiting for it given there is no published road-map of intent. ps. It would be nice to have a stable interface that wasn't LUA, eg. a comprehensive debrief log. LUA is great for scriptlets but for a project the size of a dynamic campaign it is the wrong tool. pps. yeah, c0ff is always very helpful. 'Big Ups' to him.

Hi Vekkinho, the VNAO web admin (Thunderbird, also-known-as 't-bird') has moved the site to the data center he works at and there are some 'teething issues'. I'll forward this post to his VNAO PM address.

Ah, lubey, they covered this in my Air Force flight theory two decades ago (yep, I'm getting on). The turbulent boundary layer creates more drag than laminar flow, which would ordinarily cause a shorter distance for the golf ball. However, the turbulent layer has more energy (as well as more drag) so it follows the surface for a longer distance before separating. The laminar flow is less energetic and separates sooner. So while the portion of laminar flow has less drag than turbulent flow the earlier separation results in a higher overall drag for the smooth surface (golf ball in this case). The golf ball is dimpled to introduce turbulent flow and delay separation, reducing drag and giving a longer flight. You'll see a similar trick on some aircraft (the A4 Skyhawk is the only one I saw close up). Where little strakes are put around the middle of the wing chord (as the earlier part of the chord is in laminar flow and produced most lift). The strakes also help prevent sideways movement of air, reducing wingtip 'spillover' (which modern commerical jets now have winglets for). So for a wing you want a very smooth leading edge back to around 1/3 chord (hence the "no step" signs) and after that the surface roughness doesn't matter as much.

Hey, what about a spider simulator as the next object modelled by DCS? http://www.onemotion.com/flash/spider/

You are aware that flying a real aircraft you are affected by glare. A helmet sunshield reduces it but never eliminated it, in my (limited) experience. Maybe I'm too 'hardcore', but removing sun glare removes some of the challenge. "Watch for Huns in the sun" (an old RAF saying) is meaningless if you start making detection unrealistically good. To each their own/horses for courses/strokes for folks/<insert cliche here> - you are free to change the game as you please. Just be aware that you are reducing your realism. Then there's no need for good airmanship by choosing to approach your target from a different vector that doesn't put the sun dead ahead, so a little something is lost without glare.

The reciprocal view Nice pics Kaiza, thanks for posting them. Was a great mission - the premise was hunting pirates down on the Black Sea and then hitting their shore bases. Some great low-level flying to identify which ships were pirates and which were not (randomized with the trigger system). Sundown, forming up on Kaiza: To me the sundown is starting to look like an oil painting (great work ED): Unfortunately the screenshots don't do justice to the awesomeness of the Night Vision Googles (NVG) and the way light sources flicker in them: Kaiza checks my ship for damage. Formation flying with NVG is cool: Unfortunately the screenshots also don't show the 'minty goodness' when you turn the instrument console lights up in night mode. This is not your grandad's simulator, for sure: Unfortunately I didn't get the screenshot I wanted of 'ilikepies' buddy lasing and Kaiza pickling his LGB. From my point-of-view all I could see is this IR pointer coming down through the clouds onto a target. We called it, "The Finger of God" - and whoever that finger was pointing at was gonna get smitten shortly thereafter. From an earlier mission. There is some bug where your aircraft tips up and nothing you can do will get it down. Check out the 'wheelie', kept this up for some distance and wouldn't go down:

The Simple Flight Model (SFM) aerodynamics and engine values are tabular and interpolated. What is not known are the forms of the: * Advanced Flight Model (Su-25A, Su-25T) * Blackshark Flight Model * Warthog Flight Model. Any of these models are sufficient, tabular or not. What matters is how many terms the equations have (how many effects they cater for) - which is balanced against the available time to compute for all aircraft and missiles currently in-game. People don't generally notice if tabular values are a little off (unless they are way off by a significant factor, eg the roll rates for unloaded aircraft until the FC2 patches). They notice when equation terms are missing or unbalanced (excessive roll in stall, poor recovery time, handling with missing outer wings etc).

I don't think you need a full compressible fluid dynamics simulation to have a great game. Perhaps the biggest barrier is the fact that the core engine is single threaded (plus another thread for sound). That means that the calculations must be very fast or they'll hold up everything else that has to go on. Once you can run the calculations in another thread you'll have several multiples of the time available now to do them (as then the calculations can run in parallel and just have to take the same amount of time or slightly less than *all the other stuff*). What is remarkable (if you haven't already considered it) is that the ED team have 'good enough' calculations at the moment given they must have difficult time constraints within the single thread. nb: ED are looking at additional multi-threading where they can (it's not an easy thing to put in an existing system). Edit: didn't realise 'quantum leap' was a peeve of sobek. Sorry man. Although part of me wants to slip that in future discussions ("oh noes, he dropped the "Q' bomb again!") :)

Yes I am well aware, thanks (from a former life, have a PhD in Physics, and some of that junk still remains in my head). A quantum transition is the change in any two discrete states. On the atomic electronic (electron) scale this can be the smallest change in energy state of a bound electron (free electrons are generally considered to have continuous energy states for all useful intents and purposes). The phrase "quantum leap" is a colloquially used to mean something else, and is more akin to the wavefunction having a non-zero probability on the 'far' side of a potential barrier (aka 'tunneling') that is an unexpected result from Classical (intuitive) physics. Yes, you are perfectly precise when you point out the phrase "quantum leap" is inaccurate from a physics perspective. However, from a common culture perspective I don't think there was any ambiguity in the intended meaning, and sometimes you just gotta *get down* when you talk with regular joes :). If we are going to be fussy about it, then you needn't use the (incorrect) colloquialism "anal" (shortened from "anal retentive" in slightly discredited Freudian psychoanalysis), you can use the word "particular" instead. It all depends on what you'll let slip and what gets your personal 'goat'. However, don't stop the corrections - they have interesting explanations.