2.5 Test #1 - BUGTEST REPORT THREAD

I upped sv_specspeed on the testing server to 7 to make it quicker to look round the map and stuff.

Edit: That makes it slightly harder to see if there is juddering so I will leave it at 3 for now.
When I set the server up I tried different rates and specspeed settings to get rid of the spectator judder so we can rule out it being the server as well.

 

Haven't seen anyone else post this so.....
For me gas engine is listed 5 times. It is a single grey engine that is there before researching it and when equipped on anything it seems to act like no engine is equipped.

Also....Holly shit... 2013 runs so much better then 2007.

 

I thought that was fixed? Try updating your svn and check if it's still bugged.

 

It almost feels like very light, very frequent rubberbanding, like the client isn't predicting the player's movement and only updates it when the server tells it to.

 

I was wondering if it is that the client isn't applying interpolation to the player spectator entity or whatever the game refers to it as, or if the client isn't sending updates to the server fast enough while spectating.

 

That's definitely a possibility; when I was digging around the code last night there were a few instances where some of Source's more basic network and prediction code was optimized in the 2007 Empires codebase, like in the baseentity file (and such optimizations might've broken due to other code changes in 2013) or significantly changed from Source 2007 to Source 2013 (which could've broken methods used in Empires.)

I think I'll first go over the data tables because they're relatively easy to find and deal with; they're all located in the same places in each file and none of the underlying stuff involving them should've changed between 2007 and 2013. In lieu of that, I'll have to start looking into more esoteric methods of testing, such as using two separate Steam accounts to run Empires on different and trying to run the debugger on one or both of them so I'd be able to check the state of what the server sends and the player receives. Because the spectator thing happens even with local game, it'd be easier to test that, but I wouldn't know where in the code to start. Perhaps somewhere in the gamemovement or the player files.

 

Just don't give up

I'm rooting for you

 

Go team candles!

 

I got the bugs involving player positions and kill messages fixed. I'm currently trying to fix the one involving sounds in a vehicle. If I can get it fixed by Friday evening, I'll push it out, else I'll just push out what I've managed to do and we might have a test this weekend again.

Oh, and I was right. It was the datatables. It's always the datatables.

 

I can run a test this Saturday. If people want to test on Sunday then someone else will need to do it because I won't be around until much later that night.

 

Yea, I have to catch an early train every monday now, so sundays are a no-go for me.

 

I'm staring RIGHT AT THE GOD DAMN LINE that prevents the client from hearing sounds in vehicles. I could just shoehorn a fix in by having the client *always* predict weapons if they're in a vehicle, but I'd rather not do it that way.

But seriously. WHY does the vfptr properly make the calls to every function in C_EmpTank except for that single function, where it's using the parent function that always returns false? Is it because the child function isn't const when the parent function is? Most likely, but then why does it vomit with linker errors everywhere if I make it const like it should be? For that matter, why is it const in the first place? It takes a fuckin' int and returns a god damn bool. But I still get the same errors whether I make either all the IsPassengerUsingStandardWeapons const or I get rid of const on all of them.

At this rate, I'll have to make a "Shit that was fixed with a kludge and really should be done properly at some point in the future" list.

 

And then I fixed it. We should be good enough for the next test; I fixed the issues with kill messages, player positions and now with sound in vehicles. A number of bugs reported already exist in the current version of Empires or are fairly nonconsequential, so they'll go on the backburner for now until I have the chance to sort through and prioritize them.

 

There was prediction error in spectator mode in up to ~2.3/2.4. I remember I fixed it but I don't remember what it was. It might have been one of base SDK classes and not in Empires code.

 

Thanks to Ikalx and Silk for showing up at the impromptu test.

Can confirm:

Player positions now reported.
Vehicle passenger sounds/tracers thing fixed.
Camera controls panel within spectator pretty much works, though it gets stuck on sometimes and your cursor disappears.
Going prone to crouched using the crouch key tends to cause you to teleport to the map origin position 0 0 0.
Killl messages now reported.