I need to check if there is a difference in everyone's time-in-seconds sensor to see if my idea of automatic connecting and disconnecting system works.

http://beta.scratch.mit.edu/projects/10083509/

I think what might work well is to designate one person (e.g. a VPS running a program that speaks cloud protocol) who can be gauranteed to have an accurate timer, and have that single person be responsible for timing. Alternatively, you could try peer-to-peer load-balancing of the timers.

bobbybeetest wrote:

I think what might work well is to designate one person (e.g. a VPS running a program that speaks cloud protocol) who can be gauranteed to have an accurate timer, and have that single person be responsible for timing. Alternatively, you could try peer-to-peer load-balancing of the timers.

Since I don't really have a VPS, I'm going to have to stick with p2p load-balancing. Could you (or someone else) write a script to do that for me?

so idealy if you are going with the fake user idea you need
-a computer always on
-unlimited bandwith (or ridiculously high one)

Magnie wrote:

bobbybeetest wrote:

I think what might work well is to designate one person (e.g. a VPS running a program that speaks cloud protocol) who can be gauranteed to have an accurate timer, and have that single person be responsible for timing. Alternatively, you could try peer-to-peer load-balancing of the timers.

Since I don't really have a VPS, I'm going to have to stick with p2p load-balancing. Could you (or someone else) write a script to do that for me?

I'll attempt to, anyway. Do you mind if I only do pseudo code, for the sake of the not having to drag and drop everything? (or just write the algorithm to run in JavaScript or some ther very high level language)

The days since 2000 block data is hosted on the server, rather than being based on local system settings, so it should be possible to use it to eliminate the problem. See this example.

Someone reported the same bug in my version. Did I not fully understand the problem?

LiquidMetal wrote:

The days since 2000 block data is hosted on the server, rather than being based on local system settings, so it should be possible to use it to eliminate the problem. See this example.

Someone reported the same bug in my version. Did I not fully understand the problem?

We kind of need a “seconds since 2000” block, otherwise, disconnecting players can take 24 hours. :p

Edit: Oh, I didn't realize that timer included the seconds… Now I can fix this.
Edit 2: Yeah, thanks for that discovery. Now I just need people to come on and check the time. Comment please.

In my version, I replaced the (current ) block in the stage's main script with the currentTime variable. Did you specifically leave it as seconds? Also, someone reported a multiuser on mine… So I'm not sure if the problem was fixed.

EDIT: Or was that a separate problem?
You could have players add 1 to player1 taken instead of set; then check if its 2; if so, remove one and wait randomly and whichever gets it first would take it. Deadlock problem might come up then. Lists would be useful to queue. Oh well.

LiquidMetal wrote:

In my version, I replaced the (current ) block in the stage's main script with the currentTime variable. Did you specifically leave it as seconds? Also, someone reported a multiuser on mine… So I'm not sure if the problem was fixed.

No, I forgot to replace those.

You could have players add 1 to player1 taken instead of set; then check if its 2; if so, remove one and wait randomly and whichever gets it first would take it. Deadlock problem might come up then. Lists would be useful to queue. Oh well.

Yeah, I was thinking of a queue as well, but that'll get a little more complicated.

Would it make sense to have one server for dealing with this stuff? (Someone runs project from some computer that is always on?) I know this would be inconvenient if the server was “down” ever, but it might make some of the problems easier to deal with. Or you could just wait for cloud lists.

LiquidMetal wrote:

Would it make sense to have one server for dealing with this stuff? (Someone runs project from some computer that is always on?) I know this would be inconvenient if the server was “down” ever, but it might make some of the problems easier to deal with. Or you could just wait for cloud lists.

Well, it's not as portable if someone has to always have a project running on their computer… The real point of this project was to create an automatic system that dealt with new connections and disconnections so people don't have to press a Leave/Exit button before they close the web page.

Ok; Why do you have both users testing for both users timeouts instead of just testing the other? Testing its own could be separate for server disconnect, maybe?

LiquidMetal wrote:

Ok; Why do you have both users testing for both users timeouts instead of just testing the other? Testing its own could be separate for server disconnect, maybe?

You can't really have users test themselves for disconnections… Could you explain what you mean?

It seems to me that both users test the other player and themselves for disconnection. I don't understand why it would be like that.

Users cannot test themselves for disconnection and remove from list, but can alert player if cloud variables became suddenly unavailable.

LiquidMetal wrote:

It seems to me that both users test the other player and themselves for disconnection. I don't understand why it would be like that.

Users cannot test themselves for disconnection and remove from list, but can alert player if cloud variables became suddenly unavailable.

At most the users test everyone else, they can't really test themselves other than telling everyone else that they are still “active.”

I just looked at the scripts again, and I don't see the problem I was complaining about any more. Perhaps you fixed it in between, but now, it doesn't work at all. Could you add a username tracker for the p1 and p2 slots just so we can see who it thinks is in the game? Right now noone is moving but it doesn't set me into 1 or 2 and claims its full.

LiquidMetal wrote:

I just looked at the scripts again, and I don't see the problem I was complaining about any more. Perhaps you fixed it in between, but now, it doesn't work at all. Could you add a username tracker for the p1 and p2 slots just so we can see who it thinks is in the game? Right now noone is moving but it doesn't set me into 1 or 2 and claims its full.

Just fixed that bug (hopefully). It had something to do with continually checking if the other player was disconnected or not. I'll have to do some more testing later.