[The E]

I would like to see more reasoning than "best guess" first. No offense, but cryptic hints like that aren't exactly helpful.

[Kobrar44]

Issue takes place on every build 7761-RC8 and doesn't take place on 7737 and earlier. It would mean it's one of the commits between 7737 and 7761, right?

[chief1983]

I would like to see more reasoning than "best guess" first. No offense, but cryptic hints like that aren't exactly helpful.

???

Sounds like more than a guess, he narrowed down a range of commits, and found the one in that range that best fits the problem.  Short of understanding that code (something I'm not too good at) making a test build without that change would at least show whether or not it's the cause then.  But I'm sure Valathil or someone will be weighing in on that before too long with a little more authority on that code.

[The E]

I misunderstood the post, sorry. Brain not work well while sleep-deprived.

[Qent]

So it looks like now command keys issue orders in addition to their bound functions, which is a little bit annoying because I have 7-0 bound to look left/up/back/right. Is this an intentional feature? Is there any way to toggle it back to the way it was before?

[Alan Bolte]

Sorry The E, I hadn't had time to make test builds. However, it looks like build 9165 doesn't have the collision problem, yet it does have Valathil's collision reinit.

I'm working on seeing if I can narrow this down a bit better.

One other odd thing I noticed: in HEAD and 3.6.14 one of my wingmen starts Apocalypse directly in front of me so that I can ram him if I hit the afterburners, but in 3.6.12 that's not the case. Would this be worth tracking down as well? I never got retail to work on this Windows 7 box so I don't have an easy way to compare.

[chief1983]

Damn, it does?  Kobrar had tried some tests last night, including reverting 7750, and was fairly certain that had been causing at least one collision issue I think.

[Kobrar44]

I tested 9100, because it has only the old collison system which RC8 has. 9165 has the new one, which, according to what Swifty said on IRC[not sure though], might not even use code from 7750. You would have to swich to old collision system or ask Swifty.
Here are the builds I used.
http://www.mediafire.com/?y00z5y4x300ny6v

[Alan Bolte]

I built 7768 and experienced the collision problem.

I removed Valathil's collision reinit from 7768 and was able to collide with the Mentu.

[mralexs]

I googled it, and it seems a malloc error means you ran out of memory

[The E]

I googled it, and it seems a malloc error means you ran out of memory

Theoretically, yes. Practically, modern systems will almost never fail due to that (virtual memory being available and all that), unless someone does something stupid like trying to allocate MAXINT bytes of memory.

[mralexs]

Theoretically, yes. Practically, modern systems will almost never fail due to that (virtual memory being available and all that), unless someone does something stupid like trying to allocate MAXINT bytes of memory.

Any way I can fix it?

[Goober5000]

You would need to take a look at the source code.

[Swifty]

The reason why commit 7750 derped collision detection whenever change-ship-type was used was because of these lines:

Code: [Select]

obj_remove_pairs(objp);
obj_add_pairs(objp->instance);
It's supposed to be:

Code: [Select]

obj_remove_pairs(objp);
obj_add_pairs(OBJ_INDEX(objp));
objp->instance is the index of the ship/weapon/debris etc. What the collision detection system is expecting is an object index. That's obtained through OBJ_INDEX(objp).

Fixed with commit 9234.

[Spoon]

Bumping a sticky to ask what the status of this is

[niffiwan]

RC9 is ready for release, we're just waiting for it to be uploaded / hosted (I think).  And with luck, RC9 will be final *fingers crossed*

[headdie]

cool, fingers crossed indeed