Open Tsoccerguy3 opened 4 years ago
What build of bkcommonlib? And what server version (1.16?) Paper?
Please give us the actual, exact version of both plugins (There should be a /lightcleaner version
command or similar)
Also, keep in mind that those versions are still considered unstable and such results may be the result.
Paper 1.16.1 build 74 Bkcommonlib build 807 and commit , https://github.com/bergerhealer/BKCommonLib/commit/0262617b22d9945900c0878e4a1e4a7df8ee35af Light-cleaner build 72 and commit , https://github.com/bergerhealer/Light-Cleaner/commit/6eda8d2189adf6c895b0c6991e6d93aaba4a5b89
I just verified with paper 1.16.1 #76 BKCommonLib #807 and light-cleaner 1.16.1-v1 that it does in fact work. At least, when I corrupt an area light cleaner successfully repairs it. Maybe it doesn't work only for that very specific area there?
Is it a flat world gen? Do you get the same issues on normal overworlds, if they exist? Flatworlds worked for me, but maybe it fails to detect it has sky light or some such.
did a quick test . on our server we have placed barrier blocks on a grid pattern at level 256 . We have done this to stop a bug in minecraft that cause a slight pause moving from chunk to chunk . Someone discovered that when the builds varied in height to much this pause happened . The bug still there in 1.16.1 . Light cleaner was able to clean those shadows in 1.15.2 . someone laid down a snow layer and i found that i cant remove the shadow from a snow layer
when i remove the snow i can clean the shadows .
but i see that if i issue a /snow 50 command it does not place snow under a barrier block and with the snow layer in place it can not calculate lighting for a snow layer . so cleanlight fails
Do you use fast async worldedit to cause these shadows to appear?
Yeah I got them to appear but light cleaner also fixes them just fine. Do I need to create a very extremely large area of snow until I find a spot this happens in? What caused those shadows to begin with?
It also helps to know: whats the Y-level of the snow?
The server is Sanacraft 81.19.208.15:26165
The dynmap is at http://81.19.208.15:25666/ , look on grandblue FLAT map .It was a Flat map with bedrock and 20 layers of water and we created the city in the middle . Fawe was tried out a few days ago but it happens with WE too . The shadows are random , i have seen them after some edits . I think that these shadows are created by the barrier blocks we have put at level 256 to stop the chunk lag bug . I can clean the shadows , but not from a snow layer at Y - level 20
After more testing , I found that the speed of Light-cleaner is very slow in 1.16.1 and it does not like to fix the lighting of snow layers . Doing /cleanlight world takes a lot longer now
barrier blocks may interfere with fixing the light
The slow fixing has to do with slow loading of chunks. Somehow they managed to make the loading of chunks even slower now. Perhaps lowering load concurrency in light cleaner config helps reduce this issue.
I do want to fix this problem before next release, but there's also some other 1.16 issues I have to look at. Thanks for diving deep into what exactly causes it, it'll help me a lot
@bergerkiller Maybe use PaperAPI's Async chunk stuff? It has graceful fallback methods for when those methods aren't available (on non-paper servers), but will give the speed boost when possible. All you have to do is setup your depends properly and it will handle cross-compatibility.
I don't think the speed thing is really your problem if nms, bukkit, or spigot is what introduced the latest speed issue. Of course, there's a million reasons as to why the code might be running slower, so I won't tell you how to write your code! 😄
I don't think the speed thing is really your problem if nms, bukkit, or spigot is what introduced the latest speed issue.
I did a worldwide cleaning operation and after a while it just got stuck. When I teleported to the location it was cleaning, it eventually resumed. Something isn't working right in the asynchronous chunk loading / ticketing system BKCommonLib uses, which uses the same NMS api the rest of the server uses. I can't rule out a bug on my end though.
Just a note: so far I still haven't been able to repro this main issue. I made a test area and did some tests with snow/barriers and FAWE but I couldn't get it into a state that shadows appear and stay. There's a hidden variable behind this that I haven't found yet, and I worry a full world download is about the only way to debug it.
It might be worth checking if these builds fixed any of these issues. I found that since 1.14 the heightmap it calculated was incorrect, for one, it didn't include slabs. Snow might be a similar story.
https://ci.mg-dev.eu/job/BKCommonLib/869/ https://ci.mg-dev.eu/job/Light-Cleaner/79/
I did find that there's some occasional issues with lighting fixes in a super flat world that looks similar, usually redoing the clean light command and relogging helps, but I did happen upon one I couldn't fix.
These are the best steps I have to reproduce:
Were these findings with the builds I linked up above, or prior?
Sorry to leave that part out, I'll have to double check when I'm close to my computer what build I'm on, I do vaguely remember seeing the issues go away in recent updates. But if you care to try to beat me to it those are the same steps I'm gonna follow to test.
Just figured I'd help trying to reproduce.
Well, I made some provable fixes a couple hours ago, so for sure there should be some differences this time :)
But it might be some issues you had were not caused by things that have been fixed. But a broken/incorrect heightmap definitely caused some strange behavior, as that heightmap wasn't based on what sky light hits first, but something else entirely. Like where mobs can be spawned, for example. It caused areas to light up that should be covered in shadow.
So I couldn't reproduce any the above lighting issues, but I did encounter another one (#30). That being said let's leave it to @Tsoccerguy3 to confirm if the fixes work!
Not sure if this is the same exact thing, but looks pretty similar. Running the command doesn't resolve the issue. This was discovered after I deleted some nearby plots.
And doing a manual clean nearby doesnt remove those weird sourceless lighted areas? Thats weird, never seen the plugin keep light around like that
Yeah, I've been noticing some really weird behavior in this version, but I'm thinking a good deal of this is client/server mismatch. Those spots went away after the chunks were reloaded.
Here's another example of the mismatch behavior. This is covered by barrier blocks.
After running cleanlight over it, it looks like this.
But rejoining the server returns it to being dark.
Oh with that barrier issue, try updating if you haven't, since yesterday I pushed some fixes for this. Glass/barriers/leaves were considered fully opaque and it caused some strange issues like this.
Long time no see. Tried a recent patch and noticed trapdoors aren't considered transparent.
I stumbled across another issue that only presents itself on the player's texture. There's no other evidence of lighting issues other than that the player becomes dark.
I should probably be opening new issues at this point but I don't want to spam the repo with one-line snapshot build issues
I tested LightCleaner-1.16.1-v1-SNAPSHOT.jar with BKCommonLib-1.16.1-v1-SNAPSHOT.jar , No error , but does not fix the ighting errors either