[Server] Inside The Box [Puzzles]

[Wuzzy]

At the risk of sounding arrogant:
At this point I would say what this server needs is not more features, but more boxes. Tons of 'em. :P

[sofar]

At the risk of sounding arrogant:
At this point I would say what this server needs is not more features, but more boxes. Tons of 'em. :P

That's not arrogant at all, it's a completely fair assessment of what is needed to make the server a long-term success.

One of the ways that I'm trying to help with that is by alternating between making boxes easier to make, and making a box myself once in a while, and help others do that as well. If you are one of the people who wants to build - talk to me and I'll gladly help out. We can also (with a bit of manual help) allow people to build in other peoples boxes, so if you want to build together, we can set you up for that.

[v-rob]

I hope you release the source code soon. I am positively ecstatic about making boxes, but my client always lags when I try to make a box (to the point of me teleporting to an earlier position every second and no node ever actually being registered when I place it. Once it was so bad that I was dead for 5 minutes before respawning). The source code would enable me (or anyone for that matter) making my own nodes (decorative nodes), making many boxes, and releasing them as a map on the forums. This way, I could also make a coherent storyline instead of having random, scattered, isolated boxes.

To summarize: I want to be able to make a coherent storyline using puzzle boxes without have tremendous lag, hence me wanting the source code released.

Oh, no. I just realized that I have 13 boxes to go to have the new series completed again. :/

[sofar]

The source code would enable me (or anyone for that matter) making my own nodes (decorative nodes), making many boxes, and releasing them as a map on the forums. This way, I could also make a coherent storyline instead of having random, scattered, isolated boxes.

sigh

Let me put one thing up front, for everyone who *wants* the code. I am writing this somewhat agitated as I understand what you want, but you do not understand why I see it as a very, very bad thing. Especially right now.

I am NOT releasing the code so you can release your own version of something like what we do on the server.

It won't even work. The database, the server, all of it requires days of work to get working, and it's absolutely not something that players can download and "just run". It heavily relies on non-standard lua plugins that are difficult to install and run and most users won't be capable of doing. The server requires very specific minetest.conf variables that are completely incompatible with normal local games. And this is just the start of the list of reasons why it isn't something you can ship just like that to others.

Now, I'm gonna list the valid reasons why you should be excited for the release, and if it's not in this list, please don't expect the release of the code to be of any use to you:

1. You want to help out our server and code project and improve upon it
2. You want to use very specific parts to learn how it works and/or possibly reuse some of the ideas (mech nodes, skybox, maybe the music system)

That's it. Anything else is just a pipe dream, and you should expect no cooperation from me. As a matter of fact, posts like these make me less likely to even release the code at all, because the last thing that this project needs is people to pillage it like vultures and stop contributing to the project. If that is the mentality of people then I will turn around and not release the code for a very long time.

Taking good boxes is an enormous investment of time. If you take the idea somewhere else you're splitting up the community into fractured pieces and making both those groups do double work, double the server maintenance, double the security and admin duties, and both those groups now only get part of the puzzle boxes.

If you want to make coherent boxes that form a series, then you can, and you should, and you will get your own series portal, with signs and own lobby section, on the server. Several builders have already built parts of series this way and we have already planned to do seasonal puzzles when the amount of available puzzle boxes grows a bit further (2 halloween boxes isn't enough yet).

I don't understand what your lag problem is, but it's not the problem that needs solving. The server project exists because it combines good quality code with good quality content and makes people come back for more, and encourages builders to do this as well. If anyone threatens to undermine this beneficial cycle, I will not help them. I spent way too much time on this project, and money, to see it decay in disinterest because people fork it around.

If you want to build an adventure more downloadable map, this game is NOT for you. Just package mesecons and build stuff in a map and zip that up, you don't need anything from this project.

Until there are about 200-250 puzzle boxes (of good quality) I am not convinced that I should even release the code at all. I don't think the threshold for high quality content has been met for this project to sustain itself with new builders, at all. If you want this server, code, and project survive, you have to help by adding content.

[sofar]

my client always lags when I try to make a box (to the point of me teleporting to an earlier position every second and no node ever actually being registered when I place it. Once it was so bad that I was dead for 5 minutes before respawning).

FWIW, we monitor and publish server performance, including m_lag, which is a good measure of how fast the server is handling responses. We have collected several weeks of data and it shows that server performance is really good and not the cause of client lag. E.g. look at the graphs:



even the 0.2s lag spikes are nothing that would cause your client to hang for 5+ seconds. I'm suspecting you are using a problematic wifi or even cell phone connection, but you'll have that problem on any server, sadly.

Server stats (live): http://minetest.foo-projects.org/status.html

[azekill_DIABLO]

This server has a really good connection, I never suffered any lag since I play.

[the_raven_262]

Note for box-builders: don't add too many flowing water nodes.
I have had this problem when making an underwater box, I used water source nodes only on the top of the box and in the air pockets, rest would normally be filled with flowing water.
This flowing water plays a sound to the player near it, so many of these will flood (excuse the pun) the client with data, making it to time out.

[sofar]

Note for box-builders: don't add too many flowing water nodes.
I have had this problem when making an underwater box, I used water source nodes only on the top of the box and in the air pockets, rest would normally be filled with flowing water.
This flowing water plays a sound to the player near it, so many of these will flood (excuse the pun) the client with data, making it to time out.

I've added some significant code that reduces this problem. It's helped a lot already this morning, and I'll tune it to better protect against this issue going forward. The sounds are the most CPU intensive part of the server, and this isn't something we can avoid, but I'll try and do a little better with making it not affect gameplay as much as it does right now.

[sofar]

I've modified the removal tool so that you can restrict it to remove only nodes you want.

Just like the bulk tool, shift-click nodes to add it to the restriction list. Only nodes on the restriction list will get removed. This way you can remove many nodes of a certain type without touching any other nodes. Also works in right-click removal mode when you remove a large amount of nodes.

To clear the removal restriction, shift-click while pointing at air or nothing. Then you can use the removal tool as before.

[v-rob]

Ok, I realize now why you don't want to release the source code now. Bummer. :( Oh well. I'll just have to suffer my internet connection.

my client always lags when I try to make a box (to the point of me teleporting to an earlier position every second and no node ever actually being registered when I place it. Once it was so bad that I was dead for 5 minutes before respawning).

FWIW, we monitor and publish server performance, including m_lag, which is a good measure of how fast the server is handling responses. We have collected several weeks of data and it shows that server performance is really good and not the cause of client lag. E.g. look at the graphs:



even the 0.2s lag spikes are nothing that would cause your client to hang for 5+ seconds. I'm suspecting you are using a problematic wifi or even cell phone connection, but you'll have that problem on any server, sadly.

Server stats (live): http://minetest.foo-projects.org/status.html

I do realize that this is a problem on my side, that why I specified client, not server.

[sofar]

Ok, I realize now why you don't want to release the source code now. Bummer. :( Oh well. I'll just have to suffer my internet connection.

I do realize that this is a problem on my side, that why I specified client, not server.

well, we actually had an issue with the flowing water sounds over the last day, it may entirely be that the reported m_lag value isn't actually properly accounting for lua timeslice overruns, since I can't see it in the graphs but I've certainly noticed it myself. Oops.

[Wuzzy]

I think the real fix for ambient sounds would be a good generic implementation in Minetest that mods can use, instead forcing modders to write 100s of redundant mod-based implementations.

[sofar]

I think the real fix for ambient sounds would be a good generic implementation in Minetest that mods can use, instead forcing modders to write 100s of redundant mod-based implementations.

The only real fix is to use client-side modding, because it's the ABM's on the server that are the expensive problem.

The code for the ambient sounds is fairly decent, and with the latest changes would be easy to adapt into any subgame or even minetest_game, if that is wanted. It tries really hard to create both a rich atmosphere and not overload clients. Of course, it benefits from the fact that we know what box players are in and so it therefore limits the network packets significantly, but even without that it would probably do well outside this server. It's far better than any ambience mod out there, but I'm biased ;^D

[v-rob]

Ok, I realize now why you don't want to release the source code now. Bummer. :( Oh well. I'll just have to suffer my internet connection.

I do realize that this is a problem on my side, that why I specified client, not server.

well, we actually had an issue with the flowing water sounds over the last day, it may entirely be that the reported m_lag value isn't actually properly accounting for lua timeslice overruns, since I can't see it in the graphs but I've certainly noticed it myself. Oops.

Well, that explains it! I was making a box that has a lot of water in it.

[azekill_DIABLO]

I got a simple idea (to fix the bug from entering a box and go trough the ground)

Or make the ground two block thick (it's nothing really, i think, it's the simplest manner), or make the player spawn higher. It would be very pleasant, thanks.

[sofar]

Quick update: I've been working on the scoreboard code, but progress has been limited due to time availability. I did however make some significant changes to the point system on the backend, but, unfortunately, that broke a ton of stuff. This resulted in a bunch of scoring data getting garbled. Ouch. I may revert a bunch of score data to an earlier point in time when the problem was introduced to fix the problem. Otherwise, you'd have to re-play all the boxes to get proper score data into the database, and I'm not a fan of that solution.

The result is that fortunately we're a step closer to getting the scoreboards working, which will hopefully be a nice addition for players on the server. And furthermore, we may end up having the bits of code needed for players to query and view all of their progress data.

[sofar]

Well, I did the best I could, but it's likely that some player scoring / box completion has been lost in the recovery. Especially those who have played in the past week, you may end up not getting credit for boxes you have completed previously in the upcoming player score boards. I did the best I could. We'll make it so that there will be at least a way for you to see what boxes you have, and what boxes you have not have listed as complete, so you can always redo them to get the score for those.

[azekill_DIABLO]

np, as long as you imporve the server :D

[sofar]

Player ranking is now enabled!

Every time you finish a box, we keep a record of how fast, how much damage, deaths. Together, we rank you on these score numbers against other players, and the end result is a player ranking.

For now, you can view the current top 10 players in the Score section in the lobby. The scoreboard updates automatically once every hour or so, so if you have just finished a box it will take a while before the score board is updated. Your score may also lower if new boxes come out and other players complete them before you do! Of course, if you complete them, you may get back up there. Redoing boxes may improve your ranking, but it will never lower your ranking.

In the future, we'll provide players with more detailed scoring/ranking data through the terminals.

Enjoy!

[sofar]

I hate to end today on a bad note, but I'm really disappointed in some players just swearing for no reason. The server is frequented by kids and while I'm not a purist at all, I do believe that swearing in a video game is a sign of mental incapacity of some players not being able to distinguish between important issues and unimportant issues.

[sofar]

http://minetest.foo-projects.org/rankings.html

Ranking data will be viewable on the website as well. This is the same list that you can find in-game in the score section of the lobby.

[sofar]

Box rankings are now also visible on the server! This is far more complex than player rankings, and much more volatile to player actions. A high score means generally that players rate the box high but also that it's not an easy box, since the ranking system punishes easy boxes or boxes that players give up on.

This is of course now also visible on the website as well.

[azekill_DIABLO]

It's amazingly well executed... You are giving inspiration to the whole community wth such a scoreboard!

[Wuzzy]

What's the point of more features if there are no more boxes? :-(

Also, using damage, deaths and completion speed for ranking seems pointless for a puzzle game. Especially since you can just speedrun right after you completed the puzzle. But finishing a puzzle fast in the 2nd run is a pointless achievement. This is just a meaningless statistic. The only thing mildly interesting is the number of boxes completed.
Showing the 10 “best” boxes in the scoreboard for everyone to see also WILL bias future ratings.

Frankly, I don't really think the scoreboard was really neccessary.

[sofar]

...using damage, deaths and completion speed for ranking seems pointless for a puzzle game. Especially since you can just speedrun right after you completed the puzzle. But finishing a puzzle fast in the 2nd run is a pointless achievement. This is just a meaningless statistic. The only thing mildly interesting is the number of boxes completed.
Showing the 10 “best” boxes in the scoreboard for everyone to see also WILL bias future ratings.

I'm not going to divulge the actual scoring system in detail, but no matter what score system you use, there will always be someone who complains loudly about "it's unfair!". Don't be that person.

The actual ranking is a weighed rank, and none of the variables weigh in excess of any other, and so you can only marginally increase your ranking by redoing boxes (and so can others, so it's not unfair). From what I've seen so far from the data, it seems that redoing boxes has almost no effect.

Having a weighted rank for boxes means we reward builders who create boxes that encourage certain outcomes, without getting personal, and we give builders a signal that certain boxes do really well in capturing players and that may be an inspiration.

As for meaningless statistic, that's just a not well thought out statement. Even the second and third runs are statistically significant to show the amount of time needed to complete a box. If you can shave off 99.9% of the play time in a box on the second run, then it was a really bad box, and using that time in the rating for that box will lower the rating for that box. Using that time for the player ratings means you were a bit faster than someone else, so why shouldn't it count?

As for ratings, we have had so little ratings that it's almost not a factor of significance. The score boards will actually help with that, I hope, since in the future I want to rely on ratings more to do the moderation of new boxes. Having an actual measurable amount of data come from players running boxes, we can decide in the future with all that data what the best way to do that is. And these score boards are the first step towards that.

The rating system will most likely get little changes here and there to refine it and avoid abuse. It's not completed yet.