balrok

my crashfix: http://pastebin.com/m2038ed4b

ps: it's a realy bad hack and if someone asks: you don't got it from me.. ^^

the other one is just removing an assert..

please don't spam the thread with such stuff and stay on topic.. i just posted this patch cause patch in first post got no updates.. also the patch i posted is surely nothing special and if you have an already running one you shouldn't switch to this one..

ps: cyrex you should use another os.. i think win3.11 doesn't support 64bit, which is in your case better, cause you have more than 4gb ram which can't be adressed with 32bit ^^

hello just want to note here, that we're running mangos with latest 3.1.3 commit

very stable with this patch by derex: http://pastebin.com/m7d3df483

stable means in this case no crash for 48h with 1200 user peak

actualy i had to implement two patches (but are hacks ) for some long known crashes/freeze - but don't know if they are related to the patch so i won't post them here

we could change the HandlePLayerLeaveZone a bit

cause if we leave an outdoor-zone we have already the m_OPVP-pointer set inside playerclass and can directly call the outdoor-class for the leave..

but have to go now and can't write code.. maybe this evenign i'm back

would be interesting whats on:

/usr/lib/gcc/x86_64-redhat-linux/4.1.2/../../../../include/c++/4.1.2/ext/hashtable.h:513

i don't see why a getcreature could crash.. even with a nonexistant guid this shouldn't happen

fix for first crash with bg - is now compiling and will be pushed soon

betaman2k this is no crashlog, this is just a log

use gdb to create crashlogs

i got one from the assert(false) in BuildPlayerRepop - i'm using 3.1.3 but have seen this error on 2.4.3 too .. removing the assert fixes the crash, but maybe someone is interested what's wrong there

0x00007f9b39789ed5 in raise () from /lib/libc.so.6
#0  0x00007f9b39789ed5 in raise () from /lib/libc.so.6
#1  0x00007f9b3978b3f3 in abort () from /lib/libc.so.6
#2  0x00007f9b39782dc9 in __assert_fail () from /lib/libc.so.6
#3  0x00000000006d6b1d in Player::BuildPlayerRepop (this=<value optimized out>) at ../../../src/game/Player.cpp:4067
#4  0x0000000000791cc9 in WorldSession::LogoutPlayer (this=0x7f9b34244650, Save=true) at ../../../src/game/WorldSession.cpp:302
#5  0x0000000000792bbb in WorldSession::Update (this=0x7f9b34244650) at ../../../src/game/WorldSession.cpp:275
#6  0x000000000078c368 in World::UpdateSessions (this=0x7f9b34023750, diff=248) at ../../../src/game/World.cpp:2042
#7  0x000000000078c526 in World::Update (this=0x7f9b34023750, diff=248) at ../../../src/game/World.cpp:1590
#8  0x000000000053c28d in WorldRunnable::run (this=<value optimized out>) at ../../../src/mangosd/WorldRunnable.cpp:64
#9  0x000000000082cf06 in ACE_Based::Thread::ThreadTask (param=0xd3a53d0) at ../../../src/shared/Threading.cpp:183
#10 0x00007f9b3a257fc7 in start_thread () from /lib/libpthread.so.0
#11 0x00007f9b398275ad in clone () from /lib/libc.so.6
#12 0x0000000000000000 in ?? ()
#0  0x00007f9b39789ed5 in raise () from /lib/libc.so.6
No symbol table info available.
#1  0x00007f9b3978b3f3 in abort () from /lib/libc.so.6
No symbol table info available.
#2  0x00007f9b39782dc9 in __assert_fail () from /lib/libc.so.6
No symbol table info available.
#3  0x00000000006d6b1d in Player::BuildPlayerRepop (this=<value optimized out>) at ../../../src/game/Player.cpp:4067
   data = {<ByteBuffer> = {static DEFAULT_SIZE = 4096, _rpos = 0, _wpos = 3, _storage = {<std::_Vector_base<unsigned char, std::allocator<unsigned char> >> = {
       _M_impl = {<std::allocator<unsigned char>> = {<__gnu_cxx::new_allocator<unsigned char>> = {<No data fields>}, <No data fields>}, 
         _M_start = 0x7f9b2808f200 "\\003II", _M_finish = 0x7f9b2808f203 "", _M_end_of_storage = 0x7f9b2808f209 "ª\\n(\\233\\177"}}, <No data fields>}}, m_opcode = 1172}
   corpse = <value optimized out>
   __PRETTY_FUNCTION__ = "void Player::BuildPlayerRepop()"
#4  0x0000000000791cc9 in WorldSession::LogoutPlayer (this=0x7f9b34244650, Save=true) at ../../../src/game/WorldSession.cpp:302
   guild = <value optimized out>
   _map = <value optimized out>
   data = {<ByteBuffer> = {static DEFAULT_SIZE = 4096, _rpos = 140303914989264, _wpos = 140304661142102, 
   _storage = {<std::_Vector_base<unsigned char, std::allocator<unsigned char> >> = {
       _M_impl = {<std::allocator<unsigned char>> = {<__gnu_cxx::new_allocator<unsigned char>> = {<No data fields>}, <No data fields>}, 
         _M_start = 0x7f9b2823d560 "@ß#(\\233\\177", _M_finish = 0x7f9b3abc1a4d "H\\213\\005\\034þ1", _M_end_of_storage = 0x7f9b2823d2d8 ""}}, <No data fields>}}, 
 m_opcode = 35996}
#5  0x0000000000792bbb in WorldSession::Update (this=0x7f9b34244650) at ../../../src/game/WorldSession.cpp:275
   packet = (WorldPacket *) 0x20c49ba5e353f7cf
   currTime = 6
#6  0x000000000078c368 in World::UpdateSessions (this=0x7f9b34023750, diff=248) at ../../../src/game/World.cpp:2042
   itr = {<std::tr1::__detail::_Hashtable_iterator_base<std::Pair<const unsigned int, WorldSession*>, false>> = {_M_cur_node = 0x7f9b34023750, 
   _M_cur_bucket = 0xffffffffff600424}, <No data fields>}
   sess = (WorldSession *) 0x20c49ba5e353f7cf
#7  0x000000000078c526 in World::Update (this=0x7f9b34023750, diff=248) at ../../../src/game/World.cpp:1590
No locals.
#8  0x000000000053c28d in WorldRunnable::run (this=<value optimized out>) at ../../../src/mangosd/WorldRunnable.cpp:64
   diff = 248
   realCurrTime = 1695528029
   realPrevTime = <value optimized out>
   prevSleepTime = 0
#9  0x000000000082cf06 in ACE_Based::Thread::ThreadTask (param=0xd3a53d0) at ../../../src/shared/Threading.cpp:183
   _task = (class ACE_Based::Runnable *) 0xd3a53d0
#10 0x00007f9b3a257fc7 in start_thread () from /lib/libpthread.so.0
No symbol table info available.
#11 0x00007f9b398275ad in clone () from /lib/libc.so.6
No symbol table info available.
#12 0x0000000000000000 in ?? ()
No symbol table info available.

sorry - fix is online in few seconds

applied zors changes to outdoor-branch now

edit: applied another patch from zor too.. now some functions got rewritten, to be less hacky (all functions creature/object spawning/despawning related)

yes, something like this is needed first http://bitbucket.org/kane/infinitycore/changeset/960452fc116e/

you also can provide a backtrace and hope for a real fix

lol realy?

since i still didn't find out for what reason setgoartkit was added (as the name sounds - it can change the model of a gameobject maybe) it would help much to remove wrong / not needed code, which is surrounding this unknown thing.. ^^

so i apply your patch right now.. if other people say it changes things the bad way, they can post here - then we can use a better patch

edit: setgoartkit calls SetByteValue

and setbytevalue calls ObjectAccessor::Instance().AddUpdateObject(this);

which sounds good for me )

realy nice solution then

often teleporters work with areatrigger - see in MiscHandler.cpp WorldSession::HandleAreaTriggerOpcode

don't know what you mean..

commit itself doesn't spawn anything.. those things will be handled by udb (or whatever db you use).. i've posted sql in udb-forum (and added link in one of my posts here too)

ah thx tom_rus i thought bg-marks getting a different mailtype..

with latest revision this should be fixed - i just check for mail from battlemasters - since they actualy don't send anything else than battlegroundmarks i think it's ok..

checking for item-ids doesn't sound so good cause we then need to take care about this code everytime a new bg comes out (which is quite often in last days)

no, only for an area.. and also i'm not very sure..

if not it can be done in SpellMgr::GetSpellAllowedInLocationError

should be possible with this http://wiki.udbforums.org/index.php/Spell_area

thank you, is applied

i backported it to mangos-0.12 now

hopefully everything will work..

for 0.12 you'll also need the sql from http://udbforums.org/index.php?topic=14328.0 (same sql should work)

i backported also some other battleground-commits read the commitmessage for more details..

this backported is needed so that alterac valley can also be implemented for 0.12 later (without it alterac-0.12 has to much hacks)

my solution was a bit different - i got rid of this oldZoneId variable

so everything got quite short:

 6251     if(m_zoneUpdateId != newZone)
6252     {
6253         // inform outdoor pvp
6254         sOutdoorPvPMgr.HandlePlayerLeaveZone(this, m_zoneUpdateId);
6255         sOutdoorPvPMgr.HandlePlayerEnterZone(this, newZone);
6256 
6257         SendInitWorldStates(newZone, newArea);              // only if really enters to new zone, not just area change, works strange...
6258 
6259         if (sWorld.getConfig(CONFIG_WEATHER))
6260         {

i just did a quick test and hope it still works as expected (:

no

it works like this

dbtableguid is stored in code and is connected to event

when a creature gets spawned i look if dbtableguid matches some event, if no->just spawn it

if yes, add creature to bg-instance creature/gameobject map (m_EventObjects[event1,event2]=creature/guid) and then look if event is active->spawn or not ->no spawn

and later if we want to spawn a specific event we look in we look in m_EventObjects[event1,event2] and spawn every creature and gameobject of this instance

queue is in battlegroundmgr.*

but if you don't understand c++ you're lost there^^

btw i was also lvl60 alterac valley player - did this the whole day.. ^^

For lower queue time : remove limits that players can join into max 2 separate battlegrounds.

our server solved it to open only 2 battlegrounds per day.. this is very effective for low-populated servers to reduce the waittime and the amount of closed battlegrounds due to too less players

about all other ideas: they sound interesting but don't expect to have someone who will code your ideas..

also regarding your signature.. i doubt that blizzdeveloper are looking in this forum^^

also, without this patch - i'm looking at it

1,2) placing an assert should be save, cause map is only needed for gameobject/creature finding and at this point the map must already be loaded.. also it could get rid of many if(map) checks

only at battleground-destructor i think it's possible that the map won't exist cause

i guess a bg also get's destroyed if no-one enqueues for them - or just think about the battleground-templates which just exist in a virtual way

3) the problem is, that the battleground exists before the map

ok now in mangos [8554]

as noted in the last commit, you'll need database-content (posted the extracted spawns in http://udbforums.org/index.php?topic=14328.0 )

and scriptdev2 for working spiritguides (not yet committed there)

ah i noticed that i forgot the thanks part in my patch..

so at this place a very big thanks to triply who helped with getting a good blueprint and for reviewing

and also vladimir for reviewing and discussion

indeed, nice to watch

the moment mangos switched to git was most amazing - big crash and hundreds of names appeared

opcodes are still handled by mangos - and also the opcode answer (my first plan was to move the answer into sd2 - but wasn't that good)

but i have to change in database npc_flags for spiritguides, cause prior this those flags was set through code - and without them you can't talk to spiritguides

---

for bg2db_av: it's now ready - had to make some changes which will get soon into bg2db too

also i added there the bonusweekends (untested - but wasn't a big change anyway, so shouldn't harm)