Nightly Playtests - Monday - Friday (13th - 17th Aug) 20:00 BST / 3pm EDT / 12pm PDT

Ok, so we're going to start our nightly playtest schedule. I won't be at every one but I'll try to make sure the server is up. I guess the first test will have 10-15 people in it, but hopefully we can get +5 people per test so that when we go for a pug on Saturday, we're at enough players to play proper rounds. You can use SVN or Steam content, but I don't expect a great number of changes from this point onwards with regard to code, unless we hit a major bug. Just a couple of outstanding issues that should be fixed over the course of the week, with regard to a GUI pane and seismics not hurting map buildables enough. However, do expect regular script, particle and map changes day by day.

The server details are, as always, epic-empires.co.uk:27051

Please, get the word out. The more testing we do, the sooner we can send these depots out. If we can get a lot more players involved then we may even end up with these tests turning into nightly pugs, which could be really enjoyable before the main release.

If anyone wants the rcon password to the server, and information on how to admin using only rcon, then send me a message over steam or a PM over the forums.

 

crash when getting out of com to build a ref, trickster yapping about something
https://dl.dropbox.com/u/25190212/hl2_4104_crash_2012_8_13T20_8_19C0.mdmp -- GetPredDescMap()

 

Same time Tuesday. See you all there. Thanks for showing up (those who bothered).

 

I was busy with RL, I'll try next time.

 

I'll try to get on tomorrow night but it will probably be after 9 BST if yous are still playing

 

Mondays mostly don't work for me timewise, rest of week should be ok.

 

This test was better attendancewise. Have a few changes that need to be made before the next test.

-60% damage Seismic
-75% DU
Still some message about autobalance when joining sometimes.
Crashes from Quetzacotl and Blizzerd, see crashdumps.
Commander targetting seems to prefer dead players (which can't be targeted).

The damage nerfs are vs map buildables. They can be changed using a cvar but we ideally need to get the default rate perfect.

Same time tomorrow, let's try and get more people. Spam your friends, give out access, and let's get this shit built up. If we can get an extra 4-5 people per night, then by Friday we should be playing full Empires rounds like a regular match.

 

43 seconds remaining, I'm omw!

 

Crash after exiting comm, similar to Lamer's situation, on Crossroads, tuesday's test.

-- GetPredDescMap()

 

my crash was also on xroads

 

So let's see:
*Dead people can still be targeted by targets.
*Targets seem a bit random, shouldn't they target the person closest to the one you're giving the target?
*Could not use the calculator at one point, it had full ammo but could not use it.

Otherwise, good job.

 

Two crashes on xroad while building 1) a rax 2) a vf (I was the comm but hopped out to build). The curious thing is that they occured simultaneously to Blizzerds crashes, so I assume it's not some "individual" problem of my code-copy/system.

http://dl.dropbox.com/u/69008971/hl2_4104_crash_2012_8_14T19_51_27C0.mdmp -- GetPredDescMap()

http://dl.dropbox.com/u/69008971/hl2_4104_crash_2012_8_14T19_55_11C0.mdmp -- GetPredDescMap()


Also: I dled the gcfs (this time really) half an hour before the test, and the autobalance bug happened again.

 

i was also comm, and i also got out of cv and some asshat started recycling my shit the moment i crashed in both

https://dl.dropbox.com/u/1415325/empires.rar -- engine errors in vphysics (jephir)

 

All of these crashdumps point to the following code:

physics.cpp, method CCollisionEvent::ShouldCollide

c_baseentity.cpp, method C_BaseEntity::GetClassname

It's odd that this is suddenly causing issues. Perhaps it's related to going back to hard collision?

 

Perhaps the p1entity is null or something like that?
Also, it's a print function? You could use the power of out commenting...

 

It's checked if it's null right before that code is executed. I haven't had the chance to dig deeper yet, but it's strange that this quite old code is suddenly throwing fits.

EDIT: MOOtant seems to have added this code back in June 2011, probably as part of some bug hunting. Perhaps it's fine if we remove it, but it is still strange that it's suddenly causing issues

 

Well it might not be suddenly, we never played properly on this version where we regularly respawned and shit until now, so it could have been there the entire time.

 

The entity pointer should be valid, otherwise the GetClassname() Method wouldn't have been called (inside which the crash happens) - also the entity pointer should be on the stack (as the hidden 'this' parameter for GetClassname() ) which could give a hint about which entity caused it.

Does the crash happen after calling into GetPredDescMap() ? Is the method possibly overridden in the entity class and/or does weird stuff for this specific entity ?

 

GetPredDescMap() is generated inside a macro, thankfully it's not too complex:

predictable_entity.h

The entity pointer is indeed on the stack, but it doesn't tell me anything useful. EDIT: m_PredMap is NULL in the minidump, although I can't immediately see why (p.s. I'm rubbish at debugging stuff like this)

 

Crashed on Canyon. Didn't do anything special, just ran around.

http://www.mediafire.com/?vhv2tqx5pr7kmhk -- solved (jephir) 4dc99e4b42d9e9c1f09932ea86efa706e391bbb5