windows build + cars

[zimluura]

tried to build on windows a few weeks back and failed. any idea when the windows build scripts will be fixed? don't remember the error, or exactly where it snagged. and it might be different now after i update again.

LE

this guy is smoother now. still lots of little difficulties since the only top view i have is from the older model

ATT

been pondering the idea of pushing this through quickly so there is a working car right away. then uploading the blend files with it so it could be a community modified mesh. that way - if anyone sees a problem they can just fix it. does this sound like a good idea?

[charlieg]

The open source motto of "release early, release often" applies to all parts of an open source project. If the .blend is out in the open, not only is it safer, but it can be modified by anybody thus can be, as you say, community improved. It means that, if you disappear for a bit, the car can still get worked on!

[joevenzon]

I can take a look at a windows build sometime tomorrow.

The elise looks really cool, and it looks like a great start on the audi.

[joevenzon]

I put updates that get the refactor compiling for me on windows in R2252; give it a shot. Make sure to re-run the tool and build setup scripts. The hosting has been a little shaky today but hopefully that's over.

[zimluura]

cool, compiled and running. i'll try to get all the car files updated with driver coordinates real soon.

how does the optional unified coordinate system work? from what i see in the xs, i'm guessing most cars will need to have all their positions re-input to take advantage of it. and if the coordinate system isn't version 2 we'll end up with the mass of the driver being in a different place from the model position. is that correct?

edit: the restart game confirmation button doesn't seem to work right now. and the leave race button exits the game. that will make it really really time consuming to re-position drivers. but i might be able to re-do wheel coordinates to put all cars into v2. is that a worthwhile goal?

[joevenzon]

edit: the restart game confirmation button doesn't seem to work right now. and the leave race button exits the game. that will make it really really time consuming to re-position drivers. but i might be able to re-do wheel coordinates to put all cars into v2. is that a worthwhile goal?

R2253 has the restart game button working. However, on windows it looks like both restart game and leave game are crashing VDrift. It looks like it's crashing inside the bullet functions. I'll see if I can find away around it....

[joevenzon]

With R2255 I completely stopped using the part of bullet that was causing crashes, so now everything seems to be working fine. This are of bullet was also causing crashes for some 64-bit users, so hopefully that's fixed as well.

[joevenzon]

how does the optional unified coordinate system work? from what i see in the xs, i'm guessing most cars will need to have all their positions re-input to take advantage of it. and if the coordinate system isn't version 2 we'll end up with the mass of the driver being in a different place from the model position. is that correct?

If the coordinate system is declared to be version 2, then ALL parts of the car file use the newer coordinate system that you defined. If the coordinate system is version 1, then everything uses the various coordinate systems that they used before. If you want I could change the version 1 drive model positioning coordinate system to be whatever makes it easier for you.

[zimluura]

well, i'm fine with trying to re-do all cars with the version 2 coordinate system. so long as nobody is attached to the v1 system.

i got the tc6 over to v2 and it doesn't seem like it'll too painful to just do them all. of course with some things i'll be making my best guess as to which coords get switched. so as long as the svn server keeps a backup there shouldn't be anyway it can go wrong.

so with your permission i'll start on that.


edit: trying to build_vdrift.sh and i now get
scons: command not found

ok, now it works. nevermind.

[joevenzon]

of course with some things i'll be making my best guess as to which coords get switched

Look at the two versions of the XS for an example:

R202:
http://vdrift.svn.sourceforge.net/viewvc...iew=markup

R240:
http://vdrift.svn.sourceforge.net/viewvc...iew=markup

so with your permission i'll start on that.

Yeah, go for it. :-)

[zimluura]

cool, will do!

noticed a lack of contact points in the new xs file. should those be removed them in the v2 conversion?

also the svn data i got (a month ago i think) has all grey interior textures on some cars, xs and tc6 are all i've played with so far. is that expected right now?

[joevenzon]

I know for sure at least with the XS that no one has made a fancier interior yet; I think it's a dark grey-ish black right now.

[joevenzon]

noticed a lack of contact points in the new xs file. should those be removed them in the v2 conversion?

Contact points are only used to add masses to the car, exactly the same as with the particles. In vamos contact points were actually how you defined the contact mesh (in addition to defining masses), so it's left over from that, and a lot of cars have contact points with masses of 0.05. If the contact point masses are < 1kg then just remove the contact points section, because it's unnecessary and might be confusing to people who are making new cars.

[zimluura]

did a few cars last night, and committed them.

when contact points are over 1kg, i add them as particles and erase the contact point section.

i've been handling all the positions ok i think. but i haven't been modifying the hinge section. the cars still seem to drive ok. is this a coordinate that has been affected by the v2 change?

also i can't get a practice game to even start if the xm is my car. i made a v2 file for it. but can't do any testing on driver positions so reverted that one.

[joevenzon]

did a few cars last night, and committed them.

when contact points are over 1kg, i add them as particles and erase the contact point section.

Sounds good!

i've been handling all the positions ok i think. but i haven't been modifying the hinge section. the cars still seem to drive ok. is this a coordinate that has been affected by the v2 change?

Yes, the hinge coordinates will be affected by the change the same way that mass positions are. Different hinge positions have really subtle effects so it's hard to test.

also i can't get a practice game to even start if the xm is my car. i made a v2 file for it. but can't do any testing on driver positions so reverted that one.

Ah, that's because the XM doesn't have a torque curve defined. Prior to the refactor a sort of default one was used in that case, but I haven't re-implemented that.