[Home]Olympic/Discussion

Robo Home | Olympic | Changes | Preferences | AllPages

What about a triathalon (melee, 1v1 and team tournament with the same robot)? There could also be "wrestling" where the robots fight on a small battlefield (like Albert's PointBlankChallenge?). -- Kawigi

I think I am the only active Swede. At least I won't have to fight for the leadership. =) I think it could be a good idea with thre events and a crown for each and a thriatlon crown too. The three events could be 1v1, melee and teams like Kawigi suggests. What about the tournament form? -- PEZ

Problably I'w the only one from Spain too :-) I vote for some format where luck plays a role and which is phased, so people can see the bots advace (or fall) during some days. -- Albert

Edit conflict: I like the idea, but the triathlon would need to be size restricted, or you can just write separate bots for each game, put the code in one bot, and have a selector depending on getOthers() and getTeamMates?(). We could also have grudge matches, with bots from countries that have history of not getting along fighting eachother. I think the 1 round idea would be *too* short, 10 rounds would work, 5 is probably the absolute minimum. As for the tournament form, it should be heat based, to copy the olympics. I also have a significant advantage, being from the same country as Paul. ;-) -- Tango

It's OK to write separate bots for each game and merge them into one I think. Let's not start the size restriction thing in general. Though it could be interesting with a crown for each regular weight class. At least in the 1v1 and Melee. Could be hard to write a team using nanos I guess. =) -- PEZ

Speaking for the Irish team, it sounds interesting. ;-) Should we use an Elo system or a premier-league style rating? - Jamougha

Merging bots to make an all round bot doesn't really fit with the olympics, the whole point of multi event competitors is that they have to be good at all of them at the same time, rather than being good at the one they are doing when they are doing it and then being good at the next one next time. With people, all the skills have to fit into one person, which is certainly limited, with bots there is no inherent limit (other than hardware restrictions, which are high enough never to be met), so we need to impose one. As for rating, i still think heats are best, ie. in 1v1, it's a knockout tournament, in melee, the top 2 (or whatever) go through from each battle. It's much simpler like that, when you are trying for a one-off tournament, rather than stable rankings. -- Tango

Since Robocode performance is limited by human abilities I think the same applies. It's pretty touch to be good at several games. Only Paul and ABC has it all so far. -- PEZ

Then there will be no difference between the score of the country's 1v1 bot, and the 1v1 score of the country's mixed bot, so there is no point in having both. -- Tango

Right. But you might have a different set of opponents in Thriatlon. And there's the luck factor. -- PEZ

Speaking for the Brazil team, it is surely interesting, so far i know, Brazil seems to have two active participants: me and Ratosh. I'll speak with him in order to organize this. If there is some brazilian interested to join us, please let us know! I am opening a page to receive sign ups and to serve as a discussion forum: Olympic/Brazil. I have also a question to u robocoders: Since language maybe a kind of barrier, i propose that we speak in our native languages at these pages. -- Axe

The idea of a leader, seems a good thing, since each team will need a spokesman. I also think that the elected leaders should elect someone to act like a chairman of the council. This chairman should act as a mediator, someone that would organize the ideas and the council. -- Axe

Another idea - SampleBot melee. -- Kawigi

Can you elaborate on that some? -- PEZ

Final score or % score against all the sample bots at the same time. Tron, Shadow and GlowBlowMelee are some top bots there at the moment. -- Kawigi

Way cool! I wish I knew something about melee coding! -- PEZ

I'm happy that you think so positive about this challange, but we shouldn't do too many field events.i'd prefer 5 as enought and i share not the opinion that there should be a triathlon because if you are allowed to mergin a melee, 1v1 and teambot to one single bot such a turnament doesn't seem to give much sense.
What about this rules:


please post what you think about it so that we can define our final rules . (what means that the turnament - coding for it - can begin )
Opened Olympic/Germany.
Your --deathcon

I think that is important to have each of the regular weight classes represented: nanos, micros, minis, and megas. -- Axe

And also when it would be started? I suggest that the competition start at the same date taht the Olimpic Games start (From the 13th to 29th August 2004) -- Axe

I'd like to vote against the open-source rule. Sharing code within teams, sure, but I'd rather not have the final bot source be open to everyone. If it's open source then top closed-source robocoders (this means you, Paul) are gonna be less likely to help their teams make killer bots because they won't want all their great ideas being copied everywhere. -- David Alves

If it's not open source I say forget the whole idea. And I also like the SampleBot melee idea very much. It reminds me so much about when I was new to Robocoding and ran lots and lots of such battles. It was before I settled for 1v1 of course. =) -- PEZ

Why is it important that it be open source? I think it's great that you're willing to share your code with everyone, but why not allow people to decide for themselves if they want to instead forcing people to either give away their code or not compete? -- David Alves

If it isn't open source, then we may as well just add a country page to the roborumble along the same lines as the authors page (not that that isn't a good idea, though, it might be interesting, it is just a completely different idea). I think the main aim should actually be to get a load of open source bots out there to help explain advanced methods, and suchlike, with the competition as secondary, and only there as an incentive. The more source code people can look at the more they can learn, and that is what robocode is all about. -- Tango

Righto! If a competing country can't work at their full potential because of having secretive team members then that's a trade-off they do. I think the Robocode evolution is greatly advanced by the open source initiatives of David Alves and others. Who knows what kind of robots we would have had today if Paul would have released all his bots open source? I think they would have been stronger than the ones we have today. Anyone disagrees? However, it's up to every robocoder to share or to close the lid. But I think some open source competitions is of good. -- PEZ

A question. Sweden (or any country) can compete with existing robots, can't they? -- PEZ

I would imagine that as long as they are open source, and do not use any code that is written by a someone from another country (without their explicit permission for it to be used in this competition, or if it was released under a license which is clear enough about its use in compitions of this kind), they should be admissible. -- Tango

That question was asked at /Brazil's page. My opinion is that we should accept only bots from active participants. About Sweden, come on PEZ, u are almost a team by yourself :)! About the open source, i tend to think it's a good idea, but if that means that people like Paul and ABC can't participate, i'll have to consider it better... -- Axe

I am oppening a table for these kind of votings below, please help to fill up the countries and obvious the votes. I am consulting my team about their opinion. -- Axe

Since the USA team is a lot larger than the other teams, I'd be willing to join Ireland or Portugal. My mom's family comes from Ireland (Roscommon) and my dad's parents came from Portugal (Madeira). --David Alves

You'd have to change your flag for that, would you? :) -- ABC

Nah I wouldn't change the flag I use for my personal bots, I'd make them under a different package. Maybe davidinho.*? That was my dad's portuguese nickname for me as a kid. :-) --David Alves

I have to say that i do not agree with that. I was born at Japan. My grandparents are from Portugal and Italy. So... My flag is brazilian. Choose your flag. If this is not alowed in the real Olympic competitions, i think it shouldn't be here. -- Axe



Votings:
Each team has a vote. Y-Yes ; N-No ; A-Abstain

Rules voting
Rules BrasilGermanUSASwedenIreland
Every bot has to be open source? N Y N
Only bots from active participants? Y Y N

It was easy convincing the Swedish team how to vote. =) I think Paul and ABC can participate even if it's an open source effort. Whatever secrets they have in the code of their bots they just can shut up about. (Oh, and I am extremely curious about those secrets!) -- PEZ

I got no secrets, you can always decompile my bots. I like open source, it's just that I think that when you code with open source in mind you should always make it as well organised as possible, and that is not the way I robocode ;). I don't think you can learn much from reading most of the open source bots around, not robocoding and even less java. I prefer a text explaining the principles used in a successful bot, so that anyone can have fun trying to implement them their own way. Code re-use is not very usefull in robocode, the fun part is making it yourself. -- ABC

Partly agreed. I think few things can express an idea as well as code implementing it. -- PEZ

Unless it's well structured and readable code (or very simple), not for me. The only code I fully understand is the one I write, and even that becomes hard sometimes. Besides, most of the good open source bots are mini/micro and I'm not used to using code size tricks. -- ABC

I agree with ABC in all he is saying. Perfect. I have never learned anything reading any code. The key to be a good robocoder is to understand the concept behind the ideas and then use the ideas to improve your code (bot). I have to confess that i have already had used a decompiler to watch out the DT code. It tooked about 30 minutes, then i gave up.
But ABC, that is exactly the reason that make me think is harmless to being open-source. The code is useless unless u share the ideas and concepts behind them. Come on, we need u people!! That competition would not work, unless the best join. Think about it, it is not a great deal afterall. -- Axe

As i was not able to check this site for a few days let me say some things :

  1. Because the Olympic Games are one of the hardest contests for every sportsman we shall no run this turnament for every weight class.Everyone team should feel free to let robots' code become as long as it has to be.
  2. I believe its to early for votings.At the moment there are just a few nation teams with a little number of members.Let's wait till we have more active robocoders that participate.When you see that top bots have more than 1000 downloads i can't believe that there are just so few robocoders that want to help their teams. ( we dont need just top coders but guys with cool and new ideas, too.)
  3. I agree with ABC and Axe. All bots should be open source and the best should really join this competition and there are two important reasons for that:
    1. then help others and their team to get better.
    2. they can lear.It wouldn't be a big surprise for me if within this competition there will be some completely new ideas developed about robocode and writing superb bots.in td><pinion DT will not longer be the best bot after this.

at last a personal message: if you know or are ms ( author of Ares ) please let him know about this challange and ask him to join the german team.He is the only german robocoder i know.

--deathcon

Unless this Olympics lure new and dormant Robocoders to the craft you shouldn't count on too many participants. We are a rare lot. Just look at the RoboRumble/CurrentRankings table and you'll see that only 207 bots are there and if you look at the authors ranking there are only 101. And how many of those are active? -- PEZ

But form whom else are top bots downloaded +1000 times if not from active robocoders.Or do you download your own bots twenty times a day to let it become a 'most popular download' :D --deathcon

It's an ancient bug in the repository counter. Some bots suddenly get 1000 or more downloads out of the air. It has never happened to any of my bots though. I think Marshmallow might be the most downloaded bot ever. The trick to get them downloaded often is to update them often. And if it is something I do it is to update bots often! =) I also use the download count to get a picture of how many active robocoders there are. Every new Tityus gets about 20 or so downloads in a week. And I think 5 of those might be from RR@H clients. Another way is to check the Changes page on this wiki. Granted there are som lurkers, but fact remains, we are a rare lot. I like to think of us as an exclusive club. =) -- PEZ

I'm in contact with "ms" (his name is Manfred Schuster), his most recent work has been in HaikuBots :-) I thow him a note. -- Kawigi

I agree with PEZ, i don't think that we can expect a massive signing up. I also think that the sooner we decide at least a set of basic rules, the best will be. We have to at least decide wich kind of bots would be accepted and wich competitions will we have. This should also atract more people to it. -- Axe

Just Wondering, will the competitor limit be 1 bot per Country per weight class + 1 team? or will it be 1 per country per team member? I myself believe that the former is the best choice, as some countrys barely have representation(though I think PEZ counts as 3 by himself).--Dan

I think it should be set to one per event (but the same bot could potentially be used for multiple events I suppose). Maybe two per event if that sounds more fun. -- Kawigi

Opened Olympic/Votings.If you can't find your contry just add it. --deathcon

I would be interested but the problem is I'm the only one for my country (Switzerland) and I can't code all bot needed. I saw that many people where in this case (or at least alone). What about making coalition ? Coalition would be if you don't have a team for your country only and up to three members.

And what about a RobocodeDeathMatch event ? This can be Melee and/or OneOnOne.

-- Synnalagma

You probably can interest Rozu for the prospect. -- PEZ

Any active Canucks? I know ArtOfWar? was Canadian, and HumblePieLite flies a maple leaf, but are any of us still around? If this isn't until August I'll more than likely be able to find some time if anyone else is interested. -- Kuuran

Yes,and you are not the only one trying to find people coming from the same country as you do.@Synnalagma: rz - author of GlowBlow is from Swizzerland too and i don't think coalitions are a good idea.if we would take 1 bot per country and field event you would have to code around 5 bots which i think is not too much and could be done by a single person.Why don't you open a site for your country ( not only you.if i were PEZ i would not be sure to be the only robocoder from sweden).--deathcon

By the way, I'm in some kind of contact with anyone competing in the RobocodeLittleLeague, so if people want me to send more notices around for their countrymates, I can. I also can maybe convince Tobe to join in PEZ's fight. -- Kawigi

I forgot Vuen is Canadian, silly me *throws objects at Vuen* get over here. Kawigi, if anyone you're in contact with happens to be to your knowledge Canadian I'd appreciate you sending them the way of this page :), but I can't actually think of anyone. -- Kuuran

Hmmm... Can't think of anyone right off, either in the RLL realm. Of course there's several people I'm not sure about. I'll make a nice big announcement on the RLL site for it. I'm not really sure how much traffic it gets, but it might get different traffic from here. If I were you, I'd look up Ray Vermette (author of RayBot and TheArtOfWar) and send him a note to see if he's interested in getting back into it.

One seperate idea, it appears the American team is the largest so far. Maybe it would make sense to divide east from west? That pits me against David and Jim, which might be interesting, I'm not sure who else is in the west... -- Kawigi

California, woo. -- nano

I thought David lived out West (AZ or NM). -- jim

I thought David lived in Florida or something. *High-fives nano* -- Kawigi

California baby. Instead of East / West maybe we should split the USA team into a California team and a non-California team. :-P --David Alves

I want to participate with a Stockholm team. =) -- PEZ

*Nervous laughter at suddenly feeling out numbered* I really think we should keep the US team together -- jim

i agree with jim . Most other countries have problems to find members so that they won't have to write all bots alone and you want to split your team.I don't think that this is a good idea. --deathcon

Mike Zhang, Zheng Gu, and I, Andrew Ford, will be representing Uksbekistan. -- Andrew

I think the idea is that it would be slightly unfair if the U.S. was one team, since there are so many of us. If there are no objections, I'd support California as an Olympic entrant. ;) --nano

I would like to participate as "Vasco da Gama", it's my football team. -- Axe

Nobody really WANTS to claim California as part of the USA anyways... (j/k) Of course, you can go back to the analogy to real olympics - countries with more people or simply people interested in a given sport have an advantage. -- Kawigi

When will our Olympics be? At the same time as the games at Athens? -- Alcatraz

I would generally be interested to take part in this competition. however I think it would be interesting to have some unorthodox battles (as Kawigi allready proposed a SampleBot melee or the wrestling thing are both really interesting). I mean the usual battle-styles don't necessarily need more comparison (for something we have the rumble here) and we all know who has the best bots in this categories. so the good point of unorthodox battles are that no one can estimate the results before the "olympic games" if the teams keep their developments secret. wouldn't it be funnier this way? or am I totaly wrong :) -- rozu

I think you definately have a point. And welcome back to Robocodeing! We have been missing Inde -- PEZ

rozu, what about a swiss team ? I agree with you. I thought about a team battle with Two bot that can only scan and tow bot that can only fire. Each Scanner has to be written by different team member, the same for the Shooter (that's like a real team coordination I mean between bot and between coders). It only require two coder... Synnalagma

Sure... let's fight for our country. By the way, in which language are we going to communicate. I'm from the german part. -- rozu

Hi guys, you are wonderful ... I have found about the Robocode yesterday and this idea of Olympic games sounds crazy and nice. I am from Czech repubic, programmer, but beeing new to Robocode I can hardly think about entering such a competition. But I will surely watch. BTW did you think about "where" the Robolympics will take place? If i understand it well, there is no possibility to watch the battle remotely, which implies, that the Robolympics should have a "venue". Maybe we need somebody capable running the battles and recording it to MPEG or do you thing there is otger possibility so the pleasure of watching the battle is not cofined only to the "venue" computer owner? --gorn

Haha! I guess it should be possible to set up a simple hack that allows Robocode to broadcast and spectate battles, though I don't know if anyone wants to take the time to do it. I'll look at the source a little bit and see how possible it is. The biggest problem with actually watching the battles is that they would have to be run very slowly. I will also look into the possibilty of recording battle files for playback. That should be fairly simple actually, and you could use the same format for network broadcasts. -- nano

You only need to broadcast change in position for bots and the firing events of bullets, and the initial bot positions at start of round. Everything else I can think of can be inferred from these (correct me if I'm wrong). This could be something as simple as double dX, double dY, bool fired and double bulletPower. One 32 bit packet could contain all that data. The fired state is either on or off, so one bit, then reduce each of the coordinates and the fired power to 10 bits (more than precise enough) with one bit for parity or just junk. Using bit shifts encode it all into an int and fire it off, that shouldn't be a terribly long round trip, I bet it'd easily support 30 fps. Mind you that I'm none too familiar with java's stock implementation of networking so I'd have to take a nice long read of the API before deeming this definitively simple, but I don't imagine it's too difficult. Now my question is, is this possible without a decompile using robocode's control API? -- Kuuran

Answer: definitely not. You need more than just dX and dY unfortunately, as those can only tell you a bot's heading if the bot is moving. There are also the turret and radar headings. The radar is less important, but just to be complete. I think the best way to spectate would be to use simulation and dead-reckoning so that the network can scale down transmission in the face of high traffic, and the clients won't suffer as much as they would otherwise. Data necessary include bot id, absolute x and y locations, and headings for robot, gun, and radar. Other occurances don't need dead reckoning, like gunfire and damage. A damage event must contain a bot id and the new energy level of the bot. A fire event would contain the bot id, source location, bullet power, and an accurate heading of the turret, to make sure that the client perceives the bullet travel path correctly in the face of poor simulation of the robot/turret. Robot, gun, and radar headings for dead reckoning can use bytes. The x and y locations could be 12-bits each and bit-shifted into 3 bytes. Adding a gte for the message type and another for the bot id, and that brings it to 7 bytes. At 30 fps, that would be 210 bytes/sec. Then, a bot can fire roughly 3 times per second. A fire event would be much bigger. Byte id, float x, float y, float power, float turretHeading, plus a byte for the message type, for 10 bytes, that brings us up to 240 bytes/sec. Then damage events can be received as many times as fire events, but they only need to contain a byte for the bot id and a float for the energy level, for 3 bytes, and a total of 249 bytes/sec at 30fps. Of course the dead-reckoning packets can be scaled down to probably 10fps without visible flaws, and maybe 5fps at minimum. At 10fps, it becomes 109 bytes/sec. With a 10-bot melee, that's 1090 bytes/sec per client. That's favorably comparable to modern online FPS games. If we don't care as much about reliable bullets, we could make fire event packets just a bot id and bullet power, though that could really cut down on the accuracy of the client. The long and short of it though is that it would be much easier to simply record battles and play them back. Anyone have any comments? I know a few more bytes can be shaved off; anyone wanna give it a shot? You could save a byte per packet by using deltas for the x and y coordinates and then using dead-reckoning only once per second, but I'm not sure it's worth the savings. Ohh, you could combine fire, damage, and dead-reckoning packets to cut down on overhead. Then there would be four kinds of packets: regular packets, packets with firing info, packets with damage info, and packets with both. That could cut the bandwidth down to 880 bytes/sec. Could fit up to 16 people probably on a 128k upload. -- nano

You're of course right about turret headings and bot IDs, obvious oversights on my part, sorry. But why do you need damage events at all? The viewer's Robocode is just as capable of calculating the impacts and damage as the spectator server's is, especially if you're going to give all that high accuracy data on bullets at firetime. You could get further savings by clumping all the bot changes for each tick into one packet to cut down overhead and eliminating a few things. Though in melee those would be awfully big packets, it's probably more viable for OneOnOne (the whole broadcast idea is more viable for OneOnOne). Bot IDs would only be needed on the firing packets, the combined packet could simply be in a predetermined order established by the server at match start, that's saving a byte per bot per frame. I assume your format for the reduced size floats is fixed point? In that case broadcasting the headings as radians allows you to stick almost all the bits onto the right side of the point, free precision increase (though I guess this is what you had in mind anyway). If you want to sacrifice the viewing experience a little you can eliminate radar heading tracking and only update turret heading on bullet firings (it's not too important at any other time anyway), the client could also determine bot headings from it's {x, y} deltas at a one tick delay. Heck, move the whole system to one-tick delay and the heading can be determined using the info for the next tick, resulting in current headings, and whose going to complain about a one tick lag? How do realtime audio/video streams work? It seems to me they broadcast way more data than this at better framerates than 10 fps, are they just counting on a certain amount of loss being acceptable in their case? Maybe we could do broadcasting for OneOnOne and recording for melee? -- Kuuran

How deterministic is robocode? If we manually set the starting conditions, and gave the same seed for Math.Random(), would we get exactly the same battle, or is there something i've missed? If that would work then we could just issue a controller and sma8073ata files with start conditions for each battle, and people could run them however they wanted. robotcache would have to be cleared each time, but that is easy. -- Tango

Tango. That. Is an awesome idea. I know Math.random() automatically seeds itself the first time you call it based on the current time, and it might be a different seed for each Thread...but that is something worth looking at. Kuuran, you are right that we can always increase bandwidth by increasing our tolerance for data loss, but my thoughts were that we want to see accurate battles or it's not really worth doing. The reason I included damage events is that if there is lag or some error, the client might not be able to see the bullet actually hit the robot, but if that happens, the server can just correct it. My format for headings would simply be 0...255 corresponding to 0...360. :) Also, a bot's heading can't be determined from x/y deltas if it is standing still. -- nano

The other thing you have to take into account is that some robots make Random objects that are also seeded seperately. -- Kawigi

The tolerance for data loss was a curiosity, not a suggestion. I don't really think the heading of SittingDuck is important, and most other bots move frequently enough to paint an accurate picture. As for the random seeding idea, yeah, Kawigi's voicing my initial concern about this, this can only work if we make some serious assumptions about only the default seeding being used. Plus the assumption that no other randomness is used (internal generators, instances of java.util.Random, etc). I guess if we don't use any existing bots we could outlaw anything that's inconvenient for the broadcast client...

 synchronized protected int next(int bits) {
       seed = (seed * 0x5DEECE66DL + 0xBL) & ((1L << 48) - 1);
       return (int)(seed >>> (48 - bits));
 }
Is the heart of the generator, and is completely deterministic except for seed, so on the java side as long as we all use the same seeding it should be fine. -- Kuuran

Yeah, controlling arbitrary Random objects is probably just impossible. I use one in Unnamed for gaussian random numbers. It would be possible to make a new robocode.Random class though, and have people recompile their bots using it instead of java.util.Random...but it might not be worth it. Come to think of it, any bot that used System.currentTimeMillis?() would screw things up as well. -- nano

Skipped turns could also affect the battle, and they're not even remotely predictable. - Jamougha

Maybe something is possible by recording what appears in the battleview class of RobocodeGLV014 (here there's no need to decompile anything) and since it just care about view... Then you can reuse the same battleview class to see the battle. Hope that was understable. rozu just made a page Olympic/Switzerland--Synnalagma

What do you mean by Random objects? As long as they have a seed and are otherwise deterministic, it is easy to solve. Simply make the seed a Math.Random() and that will be controlled by the broadcast seed. There is no reason to use the system time, it's just reinventing the wheel, Math.random() works fine. Already written bots would need to be checked for compatability, but new bots should be easy to ensure work. -- Tango

The problem is that you don't know how many Random objects robots will create (see java.util.Random) with different arbitrary seeds. -- [[Kawigi]%]D

As long as aWk??he seeds are derived from the one seed that is distributed, it should work, shouldn't it? -- Tango

You have no idea where the bots will be deriving these seeds from. There are plenty of deterministic sources on a machine (ala system time) that you can't predict or control in any way to ensure they're all the same. Plus that doesn't do anything about the skipped turns issue, one skipped turn can make a huge difference in some bots, for example: if Muffin skips a certain turn at the start of the round he sits still for the rest of it. Since the skipped turn issue is simply to do with the accuracy (or inaccuracy) of another system which can't be predicted, good luck. -- Kuuran

@PEZ Why don't you open a site for swedish robolympics team. Cripa? for example could join it. IMHO gorn's idea of recording the battles to MPEG is the best additional i would like streamings of all finals. --deathcon

Not enough Canadian Programmers out there - Bogeta

@Bogeta: Can't stop us from trying ;)

@Everyone: Here's how the Japanese Rumble was done: [http://www-6.ibm.com/jp/event/robocode/home/robojcup/]. Thing is, they only did one round. Have we considered that the number of rounds we run each match has to be quite low if we're recording mpegs (or DivXs? or WMVs or whatnot)? One round cost them about 1.2-1.5 megs, I made a streaming WMV at a rather good quality of Muffin vs Muffin just to test and got it down to less than half that for a round, but it's an awful format and still kind of big for 35 rounds :p

(edit) Oh, right, almost forget my biggest issue with recording: bandwidth and accessibility. We're asking someone to cough up alot of bandwidth, this isn't like PEZ running the wiki where he has to transfer a wee bit of text around and run a few perl scripts, we're asking someone to serve video to all of us. We should probably first check to make sure we have a server both capable and willing to do this somewhere in the community, and it should be volunteered, it's not fair to ask someone to make this commitment. Further, those with restrictive download caps or not on broadband will have a hell of a time viewing matches, and will probably end up only ever seeing the matches their bots participate in, which is taking away a significant portion of what the experience will hopefully be.

-- Kuuran

And what about sharing the cost bandwitch among many people. We only need some central page with links to mpeg or whatever files [i suggest making at least two quility/size versions], which could well be this wiki and than as many people as we can recrute to host 1-5 match files. These people would be revarded by a privilege of seeing them first of all others ;-) -- gorn

We could just get some free webspace of one of the many sites that offer it, and upload everything there, so no-one need use up a load of upload bandwidth. -- Tango

Most of those websites don't deal favourably with using them as a video repository. (Honestly, it is the best way I can think of, I've just never had any luck trying to host multimedia stuff on free sites so I suspect it won't work) -- Kuuran

@deathcon well, I haven't decided yet if I'm going to participate or not. It depends partly on the open source issue and partly on if I think I will have the time. You probably can't imagine how badly I want the RoboRumble crown. =) Until the only Swede has decided this I will wait with opening a section. And even if I decide to join, there's little point in talking with myself on a wiki page. =) Crippa won't join. I asked him the other day when he was over here to join a Bomberman Tournament competition. Which he won, btw.

As for bandwidth. I am not allowed to run any sort of server on my DSL line. Hosting huge video files will risk bringing the wiki and rumble servlets to my ISP's attention. Have we considered using RobocodeGLV014? I think it might be possible to make it record its drawing commands to file and then it could maybe play them back. Or something like that. That shouldn't consume too much bandwidth, should it?

-- PEZ

As far as I know, I have a *virtual* server with unlimited bandwidth (or at least very high limits). It's not doing very much at the moment, so I'd be willing to host mpegs or whatever format these videos are in for a little bit at least. -- Alcatraz

And what about using BitTorrent? ? if there's only some hours where we can download ? Just an idea... --Synnalagma

If I still have Yajags running then, I also have something around 20 GB of bandwidth per month. -- Kawigi

When are the Olympics? --Alcatraz

Anything happening here? -- Jonathan

I think this tournament has been abandoned. :-( --Vic

Well then we should start it up again. -- Jokester

Well, when are the next athletic olympics? Do we have winter olympics coming up next year? -- Kawigi

We have winter and summer Olympics, so we can have robocode olympics as a seperate one. I think they must not ne in the same time! --Krabb


Robo Home | Olympic | Changes | Preferences | AllPages
Edit text of this page | View other revisions
Last edited June 18, 2005 0:44 EST by Krabb (diff)
Search: