-
-
Notifications
You must be signed in to change notification settings - Fork 368
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Latest version of paper #2945
Comments
I think it's important problem |
I really hope that a fix is planned because I can no longer do the updates at all |
Try the alpha3 version.
|
I have the same problem with the alpha. This is so strange. Before the update, all is ok, after, problem :/ |
Try updating your java to OpenJDK.... this has solved a different issue regarding regex for other people, maybe that will work for this too. |
I have java OpenJDK and it still occurs. I can't update to alpha-3 since event-item doesn't work on 2.5 :( |
Yeah, i have a same problem. And its problem of me. |
I can give my logs if it's helpful, please help with it |
It happens to me as well. When i updated my paper from ver205 to ver210 it stopped working. |
One of the solutions found is to double the time out in spigot ... |
well, but doubling it is safe? |
Personally everything works again. Skript with its error takes longer to load but works. It takes 58 seconds and the time out is 60, which is why the server stops immediately. I don't think there are any risks |
It's the time for loading the whole server or just skript? Skript loads for ~15 seconds and then errors occurring |
set watch dog and time out to high numbers |
Completely missed this report when I checked to see if it had been reported yet so I'll close my other report (here if people want the full info I put in it: #2961 ) Issue is specifically Paper build 210 (technically 211 as 210 was a failed build) and newer, if you backdate to 209 it goes away without changing anything with Skript, your scripts or anything else on the server. |
We need to fix this, unfortunately I don't know when we have time to do that. |
For the time being on 1.15 we can at least backdate the Paper build to avoid it, but hopefully you guys have time/chance to fix it before 1.16 drops as all Paper 1.16 builds would have the issue :/ |
I don't know if it is worth saying, but with Paper-220 and 254, 1.15.2, Java 8 and Skript 2.4.1 I'm not having any issues at all, everything works fine. (on a local server on macOS, with only EssentialsX, FAWE and Skript)
|
Issue is still present on Paper 1.15.2 build 254 on my test server (using only the plugins my scripts require to lighten the overall load, as per the shorter plugins list on the report I put up on PaperSpigot's GitHub) with no changes between that setup and testing on build 209 which works fine. It doesn't happen with any/all Skript setups/amount of scripts, your setup is likely just not triggering it - however for those that it is happening to it doesn't happen if you backdate to Paper build 209 specifically. |
For those with the issue update to Paper 1.15.2 build 256 or newer - Paper have applied their "band-aid" for the issue in this build so it stops happening, though they still say Skript needs to fix this properly their end. |
Well with our fix, Skript doesn't need to do anything. I suspect Skript might actually be one of the very few plugins to justify this behavior to ensure the skripts load after ALL plugins are loaded without having to worry about depnendency hell. I'm pretty happy with the solution, but if Spigot ever also starts the watchdog on the first tick, itll be back in same boat. To explain the issue, Spigot doesn't engage watchdog on the very first tick, so the fact that Skript and other plugins delayed a task to run AFTER onEnable, caused it to run within the first tick. Paper turned the watchdog on, as it's possible for your server to lock up in first tick still..... So before, people with 3 minute long Skript loads had immunity from watchdog, and now don't. On paper at least, now plugins are init BEFORE Done and before the watchdog starts, so startup time is correctly shown in "Done" too. Skript can close this issue now though since nothing needs to be done. though I hear there is a parallel mode thats off by default.... def should look at turning that on to speed up parsing, as the reporter taking around 3 minutes to load skripts is pretty insane. |
there's support for async parsing but not parallel parsing unfortunately. maybe one day, but for now the problem is njol coded skript to use static for everything everywhere and two sections parsing at once would break a billion things. |
We could still parallelize aliases parsing though, which does take a part of the initial load. That aside, we hope for skript-parser to be finished to make parsing step parallel at this point, since the original codebase was just not made to do this. |
When I looked at SkriptParser.java, it seems pretty safe to run that in parallel.... That is the slow code, so ensuring a unique instance per parsed file (which I think is already the case) should be safe to parallelize that processing... Sooo much regex and string parsing going on there, even if you synchronize every escape out of that file, you would still be net positive by significant margins. I think it should be pretty doable to parallelize that even with current code. |
Unfortunately, individual expressions sometimes refer to global state (e.g. ScriptLoader), and I think SkriptParser uses static methods relying on global state. It is probably possible to parallerize script parsing, but that would be a ton of work. Maybe more than replacing the entire parser with something faster. Since it seems I mistakenly flagged this as a bug... |
Would it be possible to call |
Also, this issue should probably be renamed. |
Description
With the latest version of paper #210, 211, 212, 213, big bug and spam on console
https://pastebin.com/GeehQTEy
Errors / Screenshots
https://pastebin.com/GeehQTEy
Server Information
Additional Context
Thank you very much
The text was updated successfully, but these errors were encountered: