Private rooms/servers

[mollthecoder]

Discussed in #112

^{Originally posted by mollthecoder May 19, 2022}
Maybe a feature to allow people to buy private servers for the game? I could make it by myself if you're okay with the idea. I really like this idea and it'd be a blast for me to work on.

[mollthecoder]

Currently working on

[m626zNq]

If people did this they would need .env that shares all info. We cant use different databases so they would need to know coders database. So its technically very dangerous to do this as they can delete everything and etc… So lets probably not do this.

[transcental]

@mollthecoder How do you plan to handle different databases, the leaderboards etc

[m626zNq]

We can only have 1 database aka the official one @DillonB07

[ExtravagantBoost]

or have randomly generated hash keys in the program turned into a url such as a url like this https://example.com/server.js?hash=abcdef123456789

[transcental]

I know. I was seeing if mollthecoder had an idea of how it would work.

[m626zNq]

Well they would need the database url, and if they have it they can easily login to it and destroy swordbattle

[codergautam]

We can only have 1 database aka the official one @DillonB07

I never said that? Please stop saying things on behalf of me lol

[codergautam]

And I presume that private servers are self-hosted, so they wouldn't connect to the official database.

[m626zNq]

@codergautam well when I was editing swordbattle on my computer, I couldnt connect to a different server because it has a different database? so I had to set the servers to undefined?

[m626zNq]

And I presume that private servers are self-hosted, so they wouldn't connect to the official database.

Well then how would they get onto the leaderboard

[transcental]

@FishyMine They'd be hosting their own instance of swordbattle. They'd have a leaderboard for their instance, accounts for their instance etc.

[m626zNq]

@DillonB07 ah I get it now, sorry

[codergautam]

Np, but I'm not sure if they would have separate accounts and leaderboards. I was imagining it as a way you can host private games just for friends, but I guess dedicated databases would be cool too.

[Scotch101Tape]

@mollthecoder How is this going? I have a few ideas on how to implement this but wouldn't want to move forward if your are already there.

[mollthecoder]

Np, but I'm not sure if they would have separate accounts and leaderboards. I was imagining it as a way you

@FishyMine They'd be hosting their own instance of swordbattle. They'd have a leaderboard for their instance, accounts for their instance etc.

Not their own instance of swordbattle, just something they can connect to from the officially hosted swordbattle, kind of like you don't need to host your own instance of minecraft to have a minecraft server. Actually, I'm drawing a lot of inspiration from minecraft for this project.

[Scotch101Tape]

^ it would just lag and crash then you would get horrible reviews. You wanna keep one server on one process.

Maybe this is true, but where are we pulling these stats from. Like we have done 0 testing on how much it can handle as far as I can tell.

[m626zNq]

^ it would just lag and crash then you would get horrible reviews. You wanna keep one server on one process.

Maybe this is true, but where are we pulling these stats from. Like we have done 0 testing on how much it can handle as far as I can tell.

I have no idea, they will never have the normal database (swordbattle's) because they could destroy the game so maybe they could have their own database they make?? I dont know

[mollthecoder]

^ it would just lag and crash then you would get horrible reviews. You wanna keep one server on one process.

Maybe this is true, but where are we pulling these stats from. Like we have done 0 testing on how much it can handle as far as I can tell.

Running a server is very resource greedy. Because these servers are web servers, WebSocket servers, and, probably databases too if we decide to store data on the server's filesystem. And a lot of game calculations need to be done server-side instead of on user's clients to prevent cheating. Also, node.js uses JavaScript obviously, particularly the V8 engine. Here, we can't efficiently take proper advantage of CPUs multitasking capabilities.

[codergautam]

I'm not very experienced on this topic either, but I would argue that running multiple instances is much more greedy than just running multiple "rooms" on a single server.

If you look at the socket.io docs there is a concept called https://socket.io/docs/v3/rooms/
I was hoping you would find a way to somehow make a Room class that can essentially run a single room of the game with it's own players, coins, and in-game leaderboard. We can host multiple of these Rooms in a single process, and it can be scaled up or down depending on the current load (something I can do). We can also make it so that the custom Rooms don't connect to the database, so that it's fair.

If we have multiple processes for every private server, well we would need a lot of resources since we are basially running multiple instances of the game, and I'm not sure if that's worth it considering that the game is declining from it's peak. Also, it's worth noting that around 70% of the accounts are from school emails. I highly doubt that these players would actually pay real money for just being able to play with their friends for a little while. Wouldn't using in-game coins be better?

TL/DR:

Some concerns are:

1. Wouldn't running multiple processes be more greedier than running multiple "Rooms" in a single process?
2. Is this something that's worth it? Considering that the game is declining and losing players, and the fact that most of the players are children, would there be enough people to buy these private servers to keep it profitable?

[mollthecoder]

I'm not very experienced on this topic either, but I would argue that running multiple instances is much more greedy than just running multiple "rooms" on a single server.

If you look at the socket.io docs there is a concept called https://socket.io/docs/v3/rooms/ I was hoping you would find a way to somehow make a Room class that can essentially run a single room of the game with it's own players, coins, and in-game leaderboard. We can host multiple of these Rooms in a single process, and it can be scaled up or down depending on the current load (something I can do). We can also make it so that the custom Rooms don't connect to the database, so that it's fair.

If we have multiple processes for every private server, well we would need a lot of resources since we are basially running multiple instances of the game, and I'm not sure if that's worth it considering that the game is declining from it's peak. Also, it's worth noting that around 70% of the accounts are from school emails. I highly doubt that these players would actually pay real money for just being able to play with their friends for a little while. Wouldn't using in-game coins be better?

TL/DR:

Some concerns are:

1. Wouldn't running multiple processes be more greedier than running multiple "Rooms" in a single process?
2. Is this something that's worth it? Considering that the game is declining and losing players, and the fact that most of the players are children, would there be enough people to buy these private servers to keep it profitable?

That could be possible. It's worth a shot.
But another idea I had:
We could make it free and open source. That way, people can host on their own servers. Also, it would still be profitable as that could increase playtime and player base, meaning more ad revenue.

We could also combine it with in-game coins or something, and allow regular players to have their own rooms too. And for other players, they could host their own servers for better performance and more customisation.

[codergautam]

@Scotch101Tape is working on the rooms idea

And sure we could have self-hosted servers I can change the setup.js file to make it work on custom servers

[Scotch101Tape]

@Scotch101Tape is working on the rooms idea

Yeahhhh I should've mentioned that earlier but I can confirm 😅

[mollthecoder]

@Scotch101Tape is working on the rooms idea

And sure we could have self-hosted servers I can change the setup.js file to make it work on custom servers

I'm actually making a custom software from scratch that mimics a vanilla server but can be modified to do whatever.

[mollthecoder]

@Scotch101Tape is working on the rooms idea
And sure we could have self-hosted servers I can change the setup.js file to make it work on custom servers

I'm actually making a custom software from scratch that mimics a vanilla server but can be modified to do whatever.

I know it sounds a bit overkill, like reinventing the wheel, but if it keeps going the way it is right now it will be worth it.

[codergautam]

K

[Scotch101Tape]

Alright its been a while...

I am at the point in the code where all I have to do is create a frontend to the private rooms, which is fairly trivial and requires no huge changes. (I also have to merge but that's a whole different horror story 💀). Anyway, the purpose of this comment is to suggest what I have thought of and also open it up to suggestions and critiques.

My current idea for private rooms is to have an api pathway (ex: /api/new-room-link) that would give a link (ex: /private/1234abcd) to a private room. Only verified players would be able to create a room and the link would break after 15 minutes if no one has joined the room. As soon as the room is empty and cleaned-up (30 seconds after becoming empty), the link would break. By break, I mean the link would give some sort of error.

Pros:

• Easy to use
• Easy to share
• Minimal frontend/UI work needed

Cons:

• Possible running out of memory through too many links (though this is probably mitigated by the time limits)
• Only verified users

[codergautam]

I like the idea of having no coins, but I feel that it would be necessary. What if a room costed 10,000 coins, and the room would be deleted after an hour.

This way, people couldn't spam create rooms, and there would be more benefits to grinding and actually collecting coins.

[mollthecoder]

I like the idea of having no coins, but I feel that it would be necessary. What if a room costed 10,000 coins, and the room would be deleted after an hour.

This way, people couldn't spam create rooms, and there would be more benefits to grinding and actually collecting coins.

10,000 coins is a bit much.

[Scotch101Tape]

I think we could prevent spamming without coins. If we limit it to one link per account and get rid of links that are not being used, it could work. As usage of the game currently stands, the server would easily be able to handle the load. If someone tried to deliberately attack the server they would run into the problem of only one link per account. The actual private room game logic won't start until the link is visited, which reduces the overhead of distributing private rooms.

Also, a private room must always have one verified account to prevent room botting.

I think this should work, though my understanding of what the captcha guarantees may be incorrect, invalidating everything I just said 😂.

[ExtravagantBoost]

I like the idea of having no coins, but I feel that it would be necessary. What if a room costed 10,000 coins, and the room would be deleted after an hour.

This way, people couldn't spam create rooms, and there would be more benefits to grinding and actually collecting coins.

well that would be too costly, why not just limiting the amount of rooms a player could make to 1 room per player?

[ExtravagantBoost]

plus that move would just turn this game more into roblox where in order to have a server you have to pay robux monthly…

[Scotch101Tape]

Ok, so its been a while...

I kinda am really good at not finishing things (if you couldn't already tell), but I really want to start finishing stuff so I can put it on a portfolio. Because of this, I have decided to once and for all finish this project. Also swordbattle.io still seems to be doing ok with visits so it won't be going to waste.

As for development, right now what I have it 13 commits ahead and 77 (😱) commits behind the main branch. What I plan to do is take what I've learned from those 13 commits, revert them, and then pull the 77 and restart the development of private rooms. The idea is that I should be done with this PR before September 🤞.

Thanks for letting me do this in the first place and for all the input you've put into this idea!

[codergautam]

Nice yeah sorry about all the new commits, and good luck :)

Let me know if you need any help, though I don't think I changed much of the backend stuff

[Scotch101Tape]

So it's September, which is when I wanted this PR finished. Obviously I haven't but I have made good progress! After looking through the old code I was able to make slight changes on the front end without adding/removing much code. As for the backend, I decided that it needs a whole rewrite for something this deep into the code. I have been working on a PoC for a way to structure the project, which is about a day of work away from being finished. Once that is complete, I'll let y'all know. Once again thanks for the patience.

[codergautam]

Cool, thanks for the update!

[SnarkyDeveloper]

any updates really want to see this in the game

[mollthecoder]

It might happen after swordbattle V2.

[codergautam]

I do plan to release something like this after the V2 update