Ok thanks for the updates. I added a KI for that: <http://developer.valvesoftware.com/wiki/SDK_Known_Issues_List#Server:_The_client.27s_team_membership_is_pointlessly_sent_across_the_wire_twice.>
There seems to be another new crasher at work then in the latest Steam. I'm getting lots of reports of game crashes as soon as they connect and get past the client loading screen, but my patch doesn't fix it. I haven't seen this myself. At 2006/05/29 05:00 PM, Yahn Bernier wrote: >Okay, I found and fixed this issue. An internal data structure was >expanded and the engine.dll was rebuilt against the updated data >structure. However, the MOD headers don't reflect the change nor would >unrecompiled MODs which would assume the old data layout. I reverted >the structure to the old version since the change wasn't necessary going >forward. We will roll out a new engine update with the fix for this >issue shortly. And yes, this bug would have lead to MOD crashes when >using this version of the engine. > >Yahn > >-----Original Message----- >From: [EMAIL PROTECTED] >[mailto:[EMAIL PROTECTED] On Behalf Of >[EMAIL PROTECTED] >Sent: Sunday, May 28, 2006 2:24 PM >To: [email protected] >Subject: RE: [hlcoders] assert since new engine update > >The following patch fixes the bug by disabling CTeam entirely. I'll >eventually remove CTeam from my mod, as it doesn't do anything, but for >now here's a backwards-compatible fix. (Unfortunately the player that >is reporting crashers since the latest Steam update was not helped by >this fix.) > > >--- mod/src/dlls/team.cpp 2005/06/11 22:14:30 1.2 >+++ mod/src/dlls/team.cpp 2006/05/28 21:20:41 >@@ -31,8 +31,11 @@ > > int SendProxyArrayLength_PlayerArray( const void *pStruct, int objectID >) > { >- CTeam *pTeam = (CTeam*)pStruct; >- return pTeam->m_aPlayers.Count(); >+ // The RecvPropArray2 in general was seemingly broken in the May >25th-ish core engine update. >+ // The CTeam in particular was certainly broken at that time, >however, thankfully it doesn't >+ // serve any important purpose. >+ // For backwards compatibility with old client, we disable it, >rather than remove it entirely. >+ return 0; > } > > > >At 2006/05/28 04:10 PM, Yahn Bernier wrote: >>We'll have someone take a look at this with a debug engine and see what >>changed. >> >>Yahn >> >>-----Original Message----- >>From: [EMAIL PROTECTED] >>[mailto:[EMAIL PROTECTED] On Behalf Of >>[EMAIL PROTECTED] >>Sent: Sunday, May 28, 2006 12:53 PM >>To: [email protected] >>Subject: Re: [hlcoders] assert since new engine update >> >>This appears to be a bug in the core engine that the SDK has no real >>control over. >> >>If I add this extra logging: >>To the bottom of SendProxy_PlayerList: >>Msg("sending player #%d = ent num %d\n", iElement, pOut->m_Int); >> >>To the bottom of SendProxyArrayLength_PlayerArray: >>Msg("sending size %d\n", pTeam->m_aPlayers.Count()); >> >>To the top of RecvProxy_PlayerList: >>AssertMsg(0, UTIL_VarArgs("got player #%d = ent num %d\n", >>pData->m_iElement, pData->m_Value.m_Int)); >> >>To the second line of RecvProxyArrayLength_PlayerArray: >>AssertMsg(0, UTIL_VarArgs("got %d (had %d) players\n", >>currentArrayLength, pTeam->m_aPlayers.Size())); >> >>Then server side on join you get: >>sending size 0 >>sending size 1 >>sending player #0 = ent num 1 >> >>And client side you get: >>c_team.cpp (31) : got 0 (had 0) players >>c_team.cpp (31) : got 1 (had 0) players >>c_team.cpp (21) : got player #1 = ent num 1 >>utlvector.h (210) : Assertion Failed: IsValidIndex(i) >>c_team.cpp (31) : got 0 (had 0) players >>c_team.cpp (31) : got 0 (had 0) players >> >>So looks like, as originally suspected, this is a crasher issue. I >have >>had reports from one person running a release build that they're now >>seeing crasher on joining a server. It's possible this is the cause. >> >>This is the only place in the SDK where RecvPropArray2 is used, and I'm >>currently looking in to just disabling this server-side, since the >whole >>CTeam thing looks like a waste of bandwidth. The player ents already >>know what team they're on, so why pointlessly send this same info via >>another method? >> >> >>At 2006/05/27 04:59 PM, [EMAIL PROTECTED] wrote: >>>I was just investigating too. It looks like in release mode there's a >>chance of it crashing, but not certain. More evidence to support my >>ongoing theory that Valve never does debug builds. :) >>> >>>At 2006/05/27 01:13 PM, Jorge Rodriguez wrote: >>>>Tony "omega" Sergi wrote: >>>>>Is anyone else getting an assert about IsValidIndex since the engine >>update? >>>>>I've made no changes between updates, and now whenever I run, I get >>this at >>>>>map start: (when first player connects) inline T& CUtlVector<T, >>>>>A>::operator[]( int i ) { >>>>> Assert( IsValidIndex(i) ); >>>>> return Base()[i]; >>>>>} >>>>> >>>>>On: >>>>>void RecvProxy_PlayerList( const CRecvProxyData *pData, void >>*pStruct, void >>>>>*pOut ) { >>>>> C_Team *pTeam = (C_Team*)pOut; >>>>>--> pTeam->m_aPlayers[pData->m_iElement] = pData->m_Value.m_Int; >>>>>} >>>>> >>>>>The line marked with the arrow there. >>>>Yes, I get it, but it hasn't crashed anything so I haven't looked >into >>>>it yet. >>>> >>>>-- >>>>Jorge "Vino" Rodriguez >>>> >>>> >>>>_______________________________________________ >>>>To unsubscribe, edit your list preferences, or view the list >archives, >>please visit: >>>>http://list.valvesoftware.com/mailman/listinfo/hlcoders >>> >>>_______________________________________________ >>>To unsubscribe, edit your list preferences, or view the list archives, >>please visit: >>>http://list.valvesoftware.com/mailman/listinfo/hlcoders >> >>_______________________________________________ >>To unsubscribe, edit your list preferences, or view the list archives, >>please visit: >>http://list.valvesoftware.com/mailman/listinfo/hlcoders >> >> >>_______________________________________________ >>To unsubscribe, edit your list preferences, or view the list archives, >please visit: >>http://list.valvesoftware.com/mailman/listinfo/hlcoders > >_______________________________________________ >To unsubscribe, edit your list preferences, or view the list archives, >please visit: >http://list.valvesoftware.com/mailman/listinfo/hlcoders > > >_______________________________________________ >To unsubscribe, edit your list preferences, or view the list archives, please >visit: >http://list.valvesoftware.com/mailman/listinfo/hlcoders _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlcoders
