Maybe have a leak, can't do pointfile

[Komag]

Due to some recent stuff I did, I suspect I have a leak. When I dmap it gets only a little ways then crash to desktop. The same thing happened to me a few months ago and it was because I had bumped one of my brushes and opened a large leak which I fixed and the crash went away. So it's the same type of situation but I did a lot of stuff and can't find the leak.

 

So I searched and read that I can do the "pointfile" but it won't work. It gives me the error:

 

Could not open pointfile:

C:/Games/doom3/darkmod/maps/komag_city44.lin

 

I don't see any ANY "lin" files anywhere though. What's going on?

[Komag]

I found and fixed the leak, problem gone, but I'd still like to know about that pointfile lin fine deal

[AstralninjA]

...

 

[greebo]

Unless the dmap command actually reports ****** leaked *******, there is probably no .lin pointfile created by the compiler, hence nothing to be found by DarkRadiant.

 

I never had the map compiler crashing to desktop at any time - what kind of map is this you're working on?

[Komag]

It IS pretty wide spread and huge, a large city mission I'm working on roughly to stress the limits of the editor and engine and anything else. Everything is just brushes mostly, with just a few patches and lights, but no finished areas.

 

Maybe I'll test just using a tiny map and having a leak and see how it works

[Komag]

Aha, I tried a little two room map with a gap in the wall and it compiled just fine with the leak warning, then in the editor I could see that nice fat red line heading right out through the gap.

 

So I guess my own map is just too much for it to work. One thing that I have no idea whether it might contribute, I haven't done ANY optimizing yet, and have placed no visportals, so the whole entire map is affected by the leak.

[Komag]

I did some more tests and it turns out in my case that it largely had to do with the width of the map (since the pointfile line always wanted to come out and fly off to the furthest side of the map). When I cut off and shrunk some brushes on the fringes and got it under 80,000 the crashes went away and leaks are properly reported.

 

This backs up my previous tests that building huge wide sprawl that extends out past the inner four sections on the whole grid is asking for trouble. The inner sections I'm talking about are the whole area from 32768 to -32768 on the X and Y (don't know about Z but I assume it would be the same).

 

So in other words, the maximum safe boundaries to build in are within a cube in the center of the editing space that extends from [-32k -32k -32k] to [32k 32k 32k].

[greebo]

So in other words, the maximum safe boundaries to build in are within a cube in the center of the editing space that extends from [-32k -32k -32k] to [32k 32k 32k].

These should go into the wiki - I don't know if there is an article about this already (Fidcal maybe knows), but we can always create one ("Engine Limits").

 

A 32k cube is still pretty huge, this should be enough for general mapping.

[Komag]

It's actually a 65k cube, but yeah, it's absolutely plenty of room

[greebo]

It's actually a 65k cube, but yeah, it's absolutely plenty of room

Heh, yes, silly me - one should think that a physicist would be able to multiply a number by 2, but I should know better by now.

[OrbWeaver]

Heh, yes, silly me - one should think that a physicist would be able to multiply a number by 2, but I should know better by now.

 

No, physicists never have to deal with piddling little numbers like 2, they would be much better at multiplying it by 2 x 10^368 or something more manageable.

[Nyarlathotep]

Never, ever trust engineers, computer scientists, physicists, or mathematicians with arithmetic. In general, never trust arithmetic to anyone whose college math education started at the calculus level.

[jdude]

lol , don't trust me

[Nyarlathotep]

But...aren't you in marketing? o_O