SpringLobby

[AF]

Ok, I posted a big thread explaining why the idea of host sharing out the map or intgrating bit torrent into the client is a big no no.

The overhed in runnin it would be unnacceptable, the time to develop ti a huge drain, and the bandwidth needed unnavailable. It would fall apart once you break the 500kb filesize barrier. UF is much better for those sorts of things.

So basically, imagine evry feature costs $5, and a dev is given $200 a week. Only some features are expensive, and bit torrents and host downloading costs $2000. Now si that worth it for what you're getting?

Your feature request is flawed and can cause harm to lobby development by making users wonder why the devs havent implemented such a simple feature without explaining to them that it isnt simple, its a lot of work, and its fundamentally flawed.

If you want the torrenting, theres nothing stopping users runnin the map torrent in azureus or utorrent

[semi]

Feature Request:

Allow clients to download the map and mod from the host of a game.

In fact, despite what af said, I'd love to put torrent support in. The truth is I don't have time to just now.


However, I think the feature request is NOT flawed. We'd like to encourage users to participate rather than be hostile to them.

[bruno07]

Feature Request:

Allow clients to download the map and mod from the host of a game.

This would rock for LAN games :)

Could be a switch in the options:
- "Allow map/mod download requests from clients when hosting"
Maybe with a sub item:
- "Only for LAN"

On LAN you would not need to worry about restricting rate or splitting files or...

/Allan

[det]

Ok, I posted a big thread explaining why the idea of host sharing out the map or intgrating bit torrent into the client is a big no no.

The overhed in runnin it would be unnacceptable, the time to develop ti a huge drain, and the bandwidth needed unnavailable. It would fall apart once you break the 500kb filesize barrier. UF is much better for those sorts of things.

So basically, imagine evry feature costs $5, and a dev is given $200 a week. Only some features are expensive, and bit torrents and host downloading costs $2000. Now si that worth it for what you're getting?

Your feature request is flawed and can cause harm to lobby development by making users wonder why the devs havent implemented such a simple feature without explaining to them that it isnt simple, its a lot of work, and its fundamentally flawed.

If you want the torrenting, theres nothing stopping users runnin the map torrent in azureus or utorrent

You fail to understand how Open Source works. If a developer comes along that has an itch to do it, then it will be done. You also seem to be confused in thinking it is really complicated to implement or that it is about bandwidth (it is mostly about ease of use).

[clericvash]

Bandwith for torrents is hardly an issue, many people have don't have a bandwith cap on their internet, therfore sending people a map over lobby would be awesome. Of course there should be an option to turn it off.

[Agon]

But how do you want to do this with bittorrent?
A bittorrent client needs a .torrent file for getting informations about files and the tracker.
I know over dht it is possible with out a tracker and the .torrent file could be send over the lobby.
Or is it possible to use only a transfer based of bittorrent?

[clericvash]

Thinking about it the torrent idea is crap to be honest.
Plenty of programs can do it extremely well.

A direct transfer from person to person would do it if that is possible, if not then the spring engine or lobby protocol needs work to allow it.

Thinking about it i just don't see they need to put bittorrent into a lobby.

[Agon]

What the transfer library should support is:
- multi peer download for one file
- downloading pieces of the file

Without this it is not really helpfully.

[AF]

The additional overhead to the lobby doesnt justify it.

There are other programs able to do it much better.

Lobby + Integrated torrent != good end result

Its one of those things that looks good but turns out to be nightmarish when you get into it. There are far better solutions and I'd like it if you all shutup about it because its a big blackhole, it consumes time and effort and spawns endless discussions that never get anywhere, blocking actual development from taking place.

det, you dont want torrents you just want a point and click download method.

Torrents wont give you that within any sane timeframe and if you did get them it wouldnt be what you wanted anyway. There are far better solutions, and the recent campaign by you and yokozar to highlight torrents is derailing progress.

And as clericvash said:

Thinking about it the torrent idea is crap to be honest.
Plenty of programs can do it extremely well.

[FoeOfTheBee]

AF seems to have some reasonable concerns, but, may they are not insurmountable. Are any other games implementing this feature or a similar feature? I don't know of any, but if it is already implemented and useful for another game, that would make me optimistic about its utility and feasibility.

If no one has done it yet, it could still be a good idea. Someone has to be first, and there may be good solutions to any concern mentioned.

As a player, what I would like most from is a Spring+Lobby deb package that works 100% with Windows players and is updated with each new spring release.

But since you are all working on this for free, I will be happy with whatever I get, as long as it doesn't break anything!

[YokoZar]

AF seems to have some reasonable concerns, but, may they are not insurmountable. Are any other games implementing this feature or a similar feature? I don't know of any, but if it is already implemented and useful for another game, that would make me optimistic about its utility and feasibility.

If no one has done it yet, it could still be a good idea. Someone has to be first, and there may be good solutions to any concern mentioned.

The lobby used to have integrated downloading from a central repository, but that put too much bandwidth costs on them. Unknown files pays for its bandwidth cost through ads - something that cannot happen from a lobby client.

Downloading from other players sidesteps this problem (whether it's via torrent or otherwise), while retaining the usability benefits of automatic downloading within the client itself.

As a player, what I would like most from is a Spring+Lobby deb package that works 100% with Windows players and is updated with each new spring release.

But since you are all working on this for free, I will be happy with whatever I get, as long as it doesn't break anything!

That I can help you with :)

[YokoZar]

Ok, I posted a big thread explaining why the idea of host sharing out the map or intgrating bit torrent into the client is a big no no.

The overhed in runnin it would be unnacceptable, the time to develop ti a huge drain, and the bandwidth needed unnavailable. It would fall apart once you break the 500kb filesize barrier. UF is much better for those sorts of things.

What overhead? There's no reason the lobby client can't stop the torrent when the game starts, and users would obviously be allowed to cap their uploads.

Your feature request is flawed and can cause harm to lobby development by making users wonder why the devs havent implemented such a simple feature without explaining to them that it isnt simple, its a lot of work, and its fundamentally flawed.

If you don't want to do it, just say so. You don't have to act like no one can do it or that the software is somehow harmed from having a feature which improves usability.

If you want the torrenting, theres nothing stopping users runnin the map torrent in azureus or utorrent

Of course. But this is just as many extra steps as downloading from unknown files in the first place. The problem isn't getting the files, it's doing it easily and conveniently without straight downloading from a central repository.

[AF]

Your chasing a pipe dream.

Even if you implemented it and got it running it still wouldnt be feasable.

• The average host ahs 256kbps upstream which in a 2v2 that means 11KBps per player disregarding the host. That means the average map would take 30-40 minutes to transfer. Therefore it needs to be P2P to get anywhere. This si not accounting for other factors sucha s the host wanting to set aside bandwidth for browsing, or slower conenctions between users over the net.
• Such a system will never be accepted unless it's turned off by default. Users will complain about it vehemently. You'll be stuck with a handful of sharing hosts and a great heaving mass of leechers.
• A P2P system would be a lot fo work to implement and test correctly. When tis finished the users then have the option of usign a primitive torrent itnegrated in or an external program that does the job 10x better and ahs an entire production team of developers working on it, such as Azureus or utorrent.
• Should a user want a map and downloads half of it then the seeder goes away they're stuck with an unfinished download.
• Itll need to eb made clear to the user that its a torrent at which point the option of using an external program instead is needed, which would always be preferable
• There are far better alternatives to using a client based downloader, be it downloading from the host or torrents.

So your system is either going to collapse through social dynamics, or its going to be made obsolete by rival programs such as azureus or utorrent. In an dieal world if we had this system it would work, but we're not, and thigns wont turn out that way. Torrents counteract this by having http seeder sites, or simply havign hgue nubmers fo users in excess of a thousand on the tracker, but our lobby usually has 200 people in it and most of thm wouldnt want to seed map downloads, not when they can get content at 10x the speed on UF.

[YokoZar]

• The average host ahs 256kbps upstream which in a 2v2 that means 11KBps per player disregarding the host. That means the average map would take 30-40 minutes to transfer. Therefore it needs to be P2P to get anywhere. This si not accounting for other factors sucha s the host wanting to set aside bandwidth for browsing, or slower conenctions between users over the net.
• Such a system will never be accepted unless it's turned off by default. Users will complain about it vehemently. You'll be stuck with a handful of sharing hosts and a great heaving mass of leechers.

On average there are about 50 people idling in #main not playing a game. Give them each a checkbox at install time "upload my maps to other users when I am not playing a game" and, even if only 20% of them click it, you've now got 10 seeds supplying maps. Given that there are typically only 12 or so people in a game waiting for it to start, and a majority of them have the map anyway, the download demands are going to be quite small. Even if only half have the map, that's still about 16 seeders for 6 leechers.

Now, admittedly, these numbers are very back-of-the envelope, but it is not at all clear how you can reject such a system out of hand because there wouldn't be enough uploaders. Public torrent sites, after all, still work, and there they don't even have the sense of community that we do.

[*]A P2P system would be a lot fo work to implement and test correctly. When tis finished the users then have the option of usign a primitive torrent itnegrated in or an external program that does the job 10x better and ahs an entire production team of developers working on it, such as Azureus or utorrent.

We don't need anywhere near the features of other torrent programs. All we need is downloading, uploading, (optional) rate limiting, and suspending of the torrent when the game starts.

[*]Should a user want a map and downloads half of it then the seeder goes away they're stuck with an unfinished download.

So?

[*]Itll need to eb made clear to the user that its a torrent at which point the option of using an external program instead is needed, which would always be preferable

I don't see why this is necessary. Sure, we could point them to some .torrent file if they wanted to download it by hand elsewhere, but we don't need to.

So your system is either going to collapse through social dynamics, or its going to be made obsolete by rival programs such as azureus or utorrent. In an dieal world if we had this system it would work, but we're not, and thigns wont turn out that way. Torrents counteract this by having http seeder sites, or simply havign hgue nubmers fo users in excess of a thousand on the tracker, but our lobby usually has 200 people in it and most of thm wouldnt want to seed map downloads, not when they can get content at 10x the speed on UF.

I'm going to make a poll about this, actually.

[AF]

I don't see why this is necessary. Sure, we could point them to some .torrent file if they wanted to download it by hand elsewhere, but we don't need to.

Why cant the lobby prepare and open the necessary program automatically?

I think right now your best finding someone and rsearching howto implement it. Otherwise your discussing it and draining time and energy.

Show that it is indeed feasable, not through public opinion, but through development and working examples, then we can talk about it, otherwise your going to become a serious drain, and we dont need drains at such a critical period in lobby development.

[Tim Blokdijk]

Well I guess people know my position on this from previous discussions. (I'm with YokoZar on this one)
I'm more or less discarding this discussion like that discussion about not using svn for development of the site.

• Should a user want a map and downloads half of it then the seeder goes away they're stuck with an unfinished download.

We can make a central server that seeds all downloads (with some cap if needed) that way files will always be available.

[clericvash]

Look guys as AF has said there is no point talking more about it we are just going round in big circles.

Here's the main dam point as it is annoying me now - If you want to work on it, that is fine, if you don't want to that is fine too.

The future - SpringLobby has some sort of torrent file support, AFLobby doesn't, oh noeeesss!!!111one. If you want it, do it, if you don't - then don't do it, simple as hell.

Now let us go back to some meaningful discussions and put this thread back on track, thank you.

[AF]

I think we should make a list of as many existing features as possible so we can enumrate the feature lists of our lobbies for comparison.

This way we know how close we are to matching tasclients featureset and what work needs doing where.

[tc-]

I've been thinking about adding BT support for a long time but haven't had the time to check if it's something we really want to have or if there is a reasonable way to implement it. There are libraries for BT like libbittorrent.

At the moment we have lots of other features that have much higher priority than this. Map and mod downloading works, though I would like it to be easier for the users. This feature request will probably show up again in half a year or something and maybe then we have the time to implement it.

Has anyone had the time to test the hosting feature yet?

[clericvash]

How exactly have you put mod and map downloading in?