Tryskell

L2J moved to Git, and that didn't help them at all to get more contributions. Of course, teamwork becomes easier, organization is also clearer (you don't have to dig in forums to find back a 2012 diff patch which maybe dissapeared), but if you are alone, there is no synergy. It helps the distribution of your project and the forking of it. Does it suddenly enlighten people to share ? I believe than no. It only helps existing sharers. I still have a lot of contributions to commit, but let's be honest, I have to review everything because most of the time, things are bugged. I only commited ONE shared script without fixing something wrong, for exemple. It's right for every single share. People often don't test all cases. Speaking of the devil... I will probably also move aCis free revision on GitHub. We will see if people shares things or not.

I was using CMS GC until now, because default GC introduces pause time (STW effect) on high RAM usage (geoengine on), a concept which can't exist on real time server (on every GC call, the server is pausing for few ms). I decided to revert to default GC when moving to OpenJDK 11, since G1GC is now improved AND default. L2OvC is currently using latest aCis and default GC, so Sahar got an overview of RAM consumption and seems happy with the results (1 month online with no variation of RAM usage). So yes, it's a good addition and a burden which dissapears. JDK12 adds a new experimental GC, but it won't fit for real time applications like L2J, so...

What shows the dude is probably coming from aCis. Both package and even lines correspond to it. I moved latest aCis to OpenJDK 11, with no specific error. As I said, the only issue (one line) was regarding my implementation of script loader, ScriptData. We jumped from OracleJDK8 to OpenJDK 11, mostly because of licence issues. So why should I think of something which actually works ? Moreover, I never translated or tested JDK9. I moved directly to 11 - and i'll probably jump to 13 when it will go, if there is. Btw, I invite you to move to JDK11, the default GC is heavily reworked and is way more performant (G1GC). JDK12 doesn't seem to have anything useful for L2J.

Open JDK 11 works with ByteBuffer clear(), the only needed edit was around ScriptData. Moreover, Oracle JDK 11 is supposed to still have this method, heritated from Buffer : https://docs.oracle.com/en/java/javase/11/docs/api/java.base/java/nio/ByteBuffer.html

remove(Integer.valueOf( is the regular writing style, over L2J and aCis. End of story.

Until JIT does its work and saves the day, it's (probably) costier.

remove(Integer.valueOf( is normally more correct.

Java remove(int) drops by the index, not the stored value. If you .remove(636) on a Java List, you delete the 636th element of your List, going IOOBE if your List hasn't enough elements. To delete the integer 636 stored into your list, you have to pass it as Integer object.

If you find it enough, it's part of aCis. Simply extract it from free revision 370. Search on the SVN for related datapack (XML, HTMs). https://xp-dev.com/svn/aCis_public/aCis_gameserver/java/net/sf/l2j/gameserver/model/actor/instance/SchemeBuffer.java https://xp-dev.com/svn/aCis_public/aCis_gameserver/java/net/sf/l2j/gameserver/data/BufferTable.java

For the cost of a new network card, HWID is aswell.

I can burst my two eyes and die peacefully after that unexpected twist. You should maybe wait the payment before telling all those lies.

NoSuchMethodError generally comes from compilation time. Your compiled pack expects the existence of a given method, but this method can't be properly retrieved. Often due if you tried to decompile/recompile your pack thinking dezipping it would be enough to edit it. I invite you to properly compile your pack, and replace your libraries only on a SHUTDOWNED server, otherwise you can have remnants of previous stuff floating in JVM. Another possibility is your IDE doesn't link things correctly, here's a topic speaking about it (read first answer) : https://stackoverflow.com/questions/35186/how-do-i-fix-a-nosuchmethoderror

None, but multisell offers more choices, like trading multiple stuff to produce one item (recipes) or trading enchanted items for other enchanted items (unseal). Not sure why NCSoft chose to keep both systems, multisell is supposed to be an advanced buylist system.

I answered you for the knownlist part on the given topic. For the whatever debuff part, I didn't understand and you didn't explain elsewhere. You have to know than aCis knownlist handles far more objects than regular L2J knownlist system (I say L2J because L2JFrozen probably didn't touch it), because : - the radius detection is bigger (was x2 bigger on the revision you tested - which translates to 4 times - it was fixed after some L2OFF tests). Which mean you see more objects on bigger distance, Z included. - L2J IL NPC knownlist CAN'T SEE OTHER NPCs. Which means any script where a NPC has to interact with another NPC, it can't work). On aCis, it's implemented. - It's not adaptative anymore (on L2J, Player knownlist radius is based on current entity number, which varies) but is now solely based on WordRegion square content, so my solution can't create ghosts (aka objects which shouldn't appear but are still populating your knownlist). L2J still got this problem, no matter the chronicle. - If, for whatever clever reason, your L2J pack actually handles NPCs seeing other NPCs on knownlist - then try to put 2k NPCs + 1 player on the same spot, and do the same on aCis. Check memory consumption. Since L2J knownlist system saves the knownlist of each people, that's 2001 Maps filled with 2000 entities = ~4 000 000 objects on Maps. On my knownlist, I wouldn't have that problem of memory consumption. All in one, getting 1kkk FPS or 230FPS doesn't change anything for your human eye. Oh, and none ever lagged. The only legit edit would be to add Z approach, to handle WorldRegion as a 3D cube instead of 2D. It would save you some FPS on places like catacombs - but most notably on ToI - with the cost of creating billions of pointless WorldRegions which will be empty forever. You also won't be able to see high flying stuff like player mounted Wyvern anymore. PS : there is strictly no reason, except knownlist, you got more FPS in one pack and in another. You probably spawned a shitload of entities somewhere. In Queent Ant Lair, I personally I'm around 55 FPS with 78 objects on knownlists. Killing those monsters and waiting for despawn ends to 65 FPS with 4 objects. Nothing else goes in the client. So, even knownlist edit can't explain the difference of FPS. And as I'm not a client editor... Not sure what you do to pump client stats up.

I'm not sure what you speak about, once again... About compilation / installation, SweeTs made some videos specifically for aCis here : https://acis.i-live.eu/index.php?topic=78.0 (reply #4) If you downloaded sources, you can edit and compile it everytime. You only have to replace the jar and additional files you edited when you need it.

Most of the static data is moved into XML, npcs included. Not sure what you don't find on those files, customs are filtered by 50000-50999.xml. For the retail NPCs, well, a simply search over txt file (notepad++ folder search is good for that) find you a name. aCis got as vocation to be as generic as possible, therefore you won't find a lot of customs stuffs or config. They can be added manually, many ppl shared stuff over years. They notably can be found on Customs section of aCis forums, accessible by customers.

Not sure what you expected more as an answer.

You don't have enough free RAM, need 2G minimum (1.4G with geodata on, which is mandatory). Verify also your OS is a 64b one, to take in consideration the amount of RAM you got (otherwise, limited to 2.8 to 3.2G on Windows 32b). If you got enough RAM, then you didn't install JDK 64b, which allows JVM to reach bigger amount than 1.4 / 1.6G. If you got a bad computer, you can also decrease the needed RAM amount, removing geodata blocks from geoengine.properties AND decreasing the amount editing the .bat / .sh file.

A 1 min search over internet would lead to : https://stackoverflow.com/questions/15796855/java-is-not-recognized-as-an-internal-or-external-command

JDK isn't installed, or paths have been cleared out somehow.

You probably use the npcId of an existing npc. Use your own custom npc to avoid using retail HTMs.

It generally means you hit the flood protection. Another way to achieve it is to add a custom which generate an error (during EnterWorld, let's say), and the error is put silent by an empty try/catch block.

There is strictly no reason to do it, and I think RooT already keep both versions of his work, so if you want to keep it updated, see with him.

ExShowTrace then. https://github.com/gmenge/l2jgreen/blob/master/L2J_Server/src/main/java/com/l2jserver/gameserver/network/serverpackets/ExShowTrace.java

I doubt it's retail, so it's probably a custom AbnormalEffect using mouse graphics. Without any client edit, you can use ExServerPrimitive to draw Lines and Points (and by extension, any form you want). Here's an exemple used for movement debug purposes. You can eventually use L2PHX on the server and see which packet is actually sent when captcha occurs.