Skript: Still memory leak

Created on 22 Dec 2018 · 11Comments · Source: SkriptLang/Skript

Description

Local variables still causes memory leaks

Steps to Reproduce

command /test:
  trigger:
    set {_i} to 1
    set {test} to true
    while {test} is true:
      add 1 to {_i}
      loop 1000 times:
        set {_test::%{_i}%::%loop-number-2%} to random number between 1 and 999999999
        send "Ram: %total ram - free ram% MB / %total ram% MB"
      wait 1 tick

command /stoptest:
  trigger:
    set {test} to false

Server Information

Server version/platform: 1.12.2
Skript version: 2.3-beta4

bug medium

Source

Duartteno

Most helpful comment

Alright, now this is looking troubling. Edit: did quick re-release of hotfix, please check if it helped.

bensku on 23 Dec 2018

👍2

All 11 comments

ok i found it...

set {_test::%loop-number-2%} to random number between 1 and 999999999 #works

leak

set {_test::%{_n}%::%loop-number-2%} to random number between 1 and 999999999

Duartteno on 22 Dec 2018

Working as intended. You're not exiting the trigger until you use /stoptest, thus local variables are not removed. The "fix" you found relies on the script overwriting existing variables instead of creating new ones.

bensku on 22 Dec 2018

👍2

@bensku after type /stoptest RAM memory stays at for example at 2Gb and it will never decrease until the restart.

Duartteno on 22 Dec 2018

JVM does not necessarily free RAM to OS, at least not until version 12 (see https://openjdk.java.net/jeps/346). Unless the expressions you use to get information about RAM have confusing names and refer to heap instead of physical memory, this is not an issue with Skript.

bensku on 22 Dec 2018

@bensku
leak1
leak2

Duartteno on 23 Dec 2018

Alright, now this is looking troubling. Edit: did quick re-release of hotfix, please check if it helped.

bensku on 23 Dec 2018

👍2

@bensku tested, issue still exists. I have no idea why.

Duartteno on 24 Dec 2018

Cannot reproduce. I did find a (relatively unlikely) memory leak, but if was fixed by commit 237b3dd.

bensku on 31 Dec 2018

@TheBentoBox Did you find a way to reproduce this?

bensku on 2 Mar 2019

No, I just labelled it since it has been sitting for months unlabelled and with no activity, but if we're not going to do anything until someone comes back with reproduction steps then we should probably just let that be reported as a new issue and close this one.