|UNESCO World Heritage Site|
|Inscription||2017 (41st Session)|
Lag is a physical effect that affects the gameplay a lot. There are few types of lag.
Types of lag
This is the worst type of lag because of its unpredictability. No-reason lags appear and disappear when they want and nothing can't be done against them. The famous lagologist User:Gabriel says that these lags appear because of cache-misses.
Mapgen lags appear when someone is exploring new areas. These lags begin slowly but can be tough when they continue.
These lags appear because of the violation of server rules. Mesecon lags begin by construction of a laggy machine and end by a permaban. These lags can be very tough.
These lags are caused by illegal Minetest forks. Some lagologists say that mobile lags haven't been approved so far.
When there are too many players connected at once, crowd lags might appear. Kicking a few of them might help…
When players are logging in and the server shows already signs of the "Crowd lag" then the additional load from the logins will increase the lag.
This is a typical user-side lag: the internet connection drops packages and actions are sometimes execute immediantly or with a longer delay. The source can be a
shitty bad internet connection of any type or a bad wireless connection to the device. This is also sometimes called 'inter-нет'-connection.
The original lagometer used a very simple average calculation utilizing the "dtime" variable. The lagologist User:Hume2 says that the lagometer showed bare nonsenses but User:orwell didn't get to doing something about this for a long time.
The improved Lagometer now uses os.clock() instead of dtime and averages 2048 samples weighed by themselves. This lagometry appears to be more accurate and useful than Version 1.
There's currently no way to predict lags but User:Hume2 offers a reward for a valuable forecasting method.
Lagometry is a science that measures the lag. The main concern of lagometry is measuring the average lag.
Let ((a, b〉, S, μ) be a space with measure where the (a, b〉 is an interval of time and μ is a measure called lagometric mesure where μ((a, b〉) = 1. Let f be a function (a, b〉 → 〈0, ∞) measurable by μ called lagometric function. The average lag is equal to integral of f over (a, b〉 by μ.
There are two known types of lagometry: Orwellian and Humean. The currently implemented lagometer uses the Orwellian lagometry.
The lagometric measure in Orwellian lagometry is the counting measure divided by the count of responses during (a, b〉. The lagometric function is a function that returns 0 when there's no response and the response time in the points where the response came.
User:Hume2 says that this lagometry doesn't give useful information when there is one huge response time next to many small response times.
The lagometric measure in Humean lagometry is λ1/(b-a) and the lagometric function returns the time until next response or 0 in the points where the response came.
According to User:Hume2, this lagometry should give more useful information than Orwellian lagometry but it wasn't tested yet.