[Vasudan Admiral]

Remember the old 'You forgot to group those objects!' message? Well it doesn't say anything now - it pretends it converted the model (ie, says it's open in the title bar), but shows nothing in the window.

Also, due to the large number of people who came onto the boards asking about that error over the years, it'd probably be a very good idea to make that error more descriptive - it'd save a lot of hassle for newbies.

Currently doing the stress test thing again with seriously screwed up models. No crashes thus far, so I'll be making them worse next.

Edit: Ok, I think I might switch to blender to manually make messy geometry - I keep killing truespace...... surprise surprise.....

[Vasudan Admiral]

Well that was good fun, but we'll have to call it a draw. I finally managed a mesh that fails to save as POF. It imported from cob ok, but seems to be in an infinite loop during POF save. (It *might* still be going - so I'll leave it running for a while longer)

I've made an 8277 poly monstrosity of an object - packed with many editions of every type of geometry error I can think of, along with some I can't.

This is the abomination in blender:


This is the abomination in PCS2:


And this is the abomination in TS:

[Kazan]

now that is one bad model 

[Bobboau]

well the thing I was testing was only active on the import, and so long as it seemed to render properly then I guess it worked.

[Bobboau]

by the way, I also made a minor tweak to the bbox code in the BSP tree to make them a little tighter, make sure I didn't make them too tight.

[Kazan]

speaking of bboxes... one of us needs to make the engine support SLDC - i'll do it, but di you have any idea where i should look to get started

[Bobboau]

that's the shield collision stuff right?

once you have all the setup done, mc_check_shield would be were the fun would happen.

[Kazan]

did you ever find the cause of that weird floating point math error

[Bobboau]

which one?

[Kazan]

the only one i knew about was where several of the bounding variables for BSP split recursion because floating point error sentinel values

[Kazan]

do you have some changes not in trunk, like a MOI calculator or something? because i'm seeing more and more models that spin waay easily under weapons fire.

[Bobboau]

are you talking about the "infinite recursion because the bbox gets so small in one direction that floating point rounding error becomes more significant than the actual box" bug?

because that's just floating error, it's just a matter of there only being so much space to describe how small of a number we are dealing with.

and I have a few fairly minor changes not related in the slightest to MOI, mostly bug fixes and BSP bbox tweaks. from my last confrontation with the MOI I determined that there probably was no way to calculate it like V did, because there seemed to be no strong relation with the V MOI settings and known MOI values for simple objects (i.e. I used the MOI for a cube on the one cargo container that is almost a featureless cube couldn't figure out how they came to there numbers)

[Kazan]

maybe you should try to get ahold of daveb

i would, but he still refuses to talk to me

[Bobboau]

I actually doubt he'd know, he wasn't the one in charge of physics, I remember him talking about the guy who was, and he was described as a walking physics book, and this is a very specific bit of code, it would require looking directly at the pertenant section of the BSPGEN code, assuming it still exists is most likely on a CD or tape (not garenteed to be in good condition) in a storage building someware in the Champaign-Urbana area of Illinois.

[chief1983]

This doesn't help any does it?  I'm guessing you already have it, or know of it, but I just thought I would check.  It's not BSPGEN, but it might still have something useful.

[Kazan]

This doesn't help any does it?  I'm guessing you already have it, or know of it, but I just thought I would check.  It's not BSPGEN, but it might still have something useful.

in a word: no
in a sentance: PCS1's compiler is a descendant of a descendant of that very extremely ungodly buggy code

[Bobboau]

the point was we won't know how it was coded unless we look at the original V code, that is not V code.

BTW, testers, how is the whole colision hole issue in the most recent builds?

[Water]

the point was we won't know how it was coded unless we look at the original V code, that is not V code.

Is the MOI different between FS1 and FS2?

[Bobboau]

god I hope not.

but what was posted wasn't based on V code, it was reverse engineered like most everything else POF related, Garry Knudson didn't have any better an idea how MOI worked than we did, and I think he just left all the fields 0 like PCS1 did, which is probably the worst thing to have done, given how MOI is an inverse tensor or something like that, which would indicate the best default would have been an identity matrix.

[Water]

god I hope not.

but what was posted wasn't based on V code, it was reverse engineered like most everything else POF

So cob2fs  - was reverse engineered?
Edit: Ah.. just seen the author name