Honestly. Is it possible to just get the server player amount. This would fix some problems people had with the update while… meeting ROBLOX halfway to preventing bots.
I was joking about this before
Obviously not exactly like this. It would be nice just to have images of the accounts without linking to their account?? Or maybe just saying 24/24 or 23/24
Another issue I thought of last night was those free robux places, I’m sure lots of users are tricked by their misleading yet accurate-looking leaderboards, and it was pretty easy to debunk when you realize that you are in a server alone. Now its harder to do that.
Now we just need to work together and create a spam bot that abuses this update
edit: seems like this post got flagged. obviously was just sarcastic poke
One of my main issues is that the group I develop at requires the ability to see all servers so that they can shut down empty ones to prevent people from joining empty servers created randomly without other servers being full and so that people can hit the “Join” button on a server so they won’t end up in a new server and all go into the place they should be at.
Posted this in a different forum thread but here’s my suggestion at the moment:
My proposal for this update is that if this update is urgently needed to prevent the spam/phishing bots, then they should at least make it so only verified (and/or at least 1+ week old or so) accounts can view servers that non-friends are in. That will cut down on a lot of the phishing bots, especially due to the fact a lot of them aren’t even verified, I’m assuming. To remove the non-friends server list overall from the whole roblox population though, is quite absurd/unnecessary.
I generally used the server list to allow people to join specific servers from my website. I allowed this for various things, for example, my site allows players in-game to communicate directly with “Server-Mods” (Other people that were given access to moderate the game by the game owner). If needed that server mod could join that server to continue assisting the user.
Is this still possible? If not how could I get around this? Possibly using the follow player endPoint?
In that case you probably know the JobId, so you could just TeleportToPlaceInstance (or even use GetPlayerPlaceInstanceAsync to get the JobId), as you already kind of know the server you want.
obviously won’t work if you don’t know the (exact) name of the caller or the server JobId it came from
This update was not meant to stop botting. It was because the server list was straining ROBLOX’s backend. Would have been nice to improve the feature so it didn’t instead of removing it for all but a select few, but oh well.
The resource issue was a secondary concern according to the OP.
Publicly-facing maybe, but through private discussion that doesn’t appear to be the case. You can put together the pieces with just what’s been said in this thread even, which is why I’m comfortable mentioning it (e.g. dodging of both the remarks that this doesn’t really help botting and the suggestions which improve upon the effectiveness against botting but reinstate the server list (bringing back the resource problem))
Hello everyone,
Please remember this post is in the public section of the Developer Forum. Do not post ways users can bot on the Roblox website since this knowledge can be used maliciously. Instead please message the ways you found to @Caelestene directly or post in the Private Bugs and Exploit Reports section of the Developer Forum… Thank you everyone for expressing your thoughts on this matter and please remember to follow our rules. If you have any questions or concerns feel free to message us directly.
Thank you,
Developer Relations Team
I’m curious how this is better for the backend. Now when someone asks for the server list, not only do they have to go request all servers, now they also need to check if you have a friend in them. Surely that should be a much bigger strain than before? The only performance increase I can think of is not having to send 20 servers over the internet, but more like an average of one (depending on #friends, #servers, …) but I doubt that would matter that much.
With it being friends only, they can probably store the current server a user is in by UserId, and then all you have to do is loop through friends and return the servers they’re in. But you’re right – I’m curious how this is substantially better than just loading 10 servers every time someone manually navigates to the game’s server tab. I’ve been curious as to whether this is a resource problem through normal usage of the API or because bots were using it so heavily (in which case #friends is a lot better than the bots spam-requesting them)
That being said, I can’t accurately answer either of these (why w/ friends is less straining and whether resource problem is b/c normal usage or botted usage) – you’d have to ask @Caelestene.
I had to guess at most of the data, but it’d seem that basically the new system does reduce the amount of network traffic for requesting servers - unless you have a lot of friend and they’re all online playing the exact same game you want to play… Then it starts suffering…
Just to clarify, I ran my UserID (2469556) through a Text-to-Binary converter, and it came out with 7 bytes (8 bits == 1 byte),
As for the server cost, I have no information as to what cost there is to requesting server information, so it’s just a guess.
The Average friend count is based off the max limit for friends set months ago (I haven’t been online in a long while - so it may have been changed).
In purple you can see the calculation for both equations - they’re both super simple, and are probably missing out a lot of information, but it does seem the new system is cost effective.
I tried to keep the averages as generic as possible, and just to note - this is suppose to be the cost of One user gong to a game page, requesting the server list.
In the new system, you also have to account for checking for every player in a server if they’re a friend. (it’d be more like 95%, as for “friend servers” you can stop the moment you find a friend). That would at least require to also load the friendlist of the requesting player.
Another thing is that the “calculations” for checking if a player is a friend also costs resources, especially for games with lots of players that you have to check.
One thing the new system doesn’t have to do, what is probably quite heavy, is getting all the avatars for all the players in all the servers *(on the current page), as now it only has to do that for players in “friend servers”
Oh, that sounds a backwards way of doing it - I thought your friends list was scanned, then if someone is in the same game, it would show up that server?..
This would certainly be a lot cheaper than scanning the current server list, and trying to find a few friends.
Ah, that could be true (for non-owners)
@Caelestene the system doesn’t seem to work 100% yet:
I just joined PF for a minute (for testing purposes) and got put in a server with a friend. The server doesn’t show up on the serverlist, except when I’m also in the server:
-- server not visible
join()
-- server visible (and shows my friend) (probably just because I'm in it)
leave()
-- server not visible again
I actually friended that person a day or two ago in the same game, if that matters.
Maybe a cached friendlist issue? no idea
I have no idea about the API or backend that goes into it, I just saw the debate as to whether it was cost effective or not, so I gathered what information I could to add to the confusion… I mean debate.
Anyhow, that’s my two pennies.
I just hope it gets sorted soon - I know I used to use the server list a lot when developing and testing my games.
Saying that, do game owners have the ability to shutdown servers now? In case of a hacker or bug?
Seems like for non-group games the place owner just has the old serverlist. Nobody can get the full list for group games for now, but they’ll add that soon ™