bind "snap to grid" to the space bar and every time you mess about with the vertex tool hit space. It has saved me a hell of a lot of trouble.

That is if you're using Trenchbroom, and also make sure you enable force integer plane points. 

I needed a door to use as a long, extending walkway. I needed it to slide out about 256 units, then up 16 units. I thought I could do it with a secret door somehow but I never could get it to cooperate. I ended up just using a func_train and used a little door hidden in the floor to provide the sound.

e4m1 actually has this too. i think that that is the one thing a secret door can't do. 

it's e2m1. 

Means you won't have to do it by hand (spacebar), unless you've already created so many degenerate brushes that you crash the editor when trying to do a global snap to grid.

But you guys won't be anywhere near that yet. 

I once snapped a perfectly fine brush to grid and it offset it by about 0.00403434 units 

that's not true, I force integer plane points and get off-grid vertices all the damn time if I don't also snap to grid.

OTP,
TB can be a little temperamental. Sometimes I mash the force to grid key a bunch of times just to make it "stick". I'm sure I'm being superstitious with that though.

Sometimes I will end up with a random leak after a while in a map to find that a brush that caused me no trouble before has suddenly degenerated and become off grid. It's not too often this happens now but I haven't tried to make any crazy terrain in a while.

My advice is make complex terrain as your last thing. 

I knew the force-to-grid advice would follow.
Reason I asked after they already were. 

when you clip a brush at lots of wacky angles, often the editor can't snap all the verts to the grid, because to snap one means slightly altering the angle of a plane to accommodate it, which would move several other verts by a different amount. this is because the verts don't determine where the planes are, the planes determine where the verts are by their intersections.

build your rocks and terrain out of brushes that are more ... brushy. make sure you clip them at integer ratio angles so that the planes always intersect at points that have a chance of being on the grid in the first place :) 

I always just use brushes that have triangle faces and move the vertexes as needed keeping them on grid. I almost never use clipping. 

Thanks Lun for your explanation.

I just saw the brushfaces slowly change in the editor and started again untill I had the right one.
Then, while mapping on, the sides became louzy, although I hadn't changed anything.
So going back to the right map slowed awfully down.
Now all brushes that stand on grid in editor start blacking out.

So editor snaps all verts untill I believe they do and then start degenerating.

@Ricky- all brushes in the screenshot are triangles that have moved vertexes, but they start telling lies. 

Wouldn't anglesense help here? 

When I compile the map without light there's just a smooth surface,
so the error isn't the misfitting brush.
I could try anglesense, but there are many parts here so many extra lights.
Replacing them with brushes for light sense I never encounterd. 

what about making the entire hill a func_wall? 

those are some good tips sock, but how am I supposed to learn how to do lighting properly?

I would recommend you find a map you like the look of (light wise) and then check out the source map file to see how it was done. Break the example map down into a small sections with light entities and gradually remove the lights one by one until you see what difference it makes.

Learning lighting for old BSP compile engines is the most time consuming thing because it involved compiling and waiting for the results. The best advice is start with lots of small test maps and try to find some good source map examples. 

WARNING: Mixed face contents (Sky, Empty) near (x y z)
WARNING: Mixed face contents (Sky, Empty) near (x y z)

I have like 20 of those when running txqbsp. 

When I compile the map without light there's just a smooth surface,
so the error isn't the misfitting brush.

the bsp structure around all those brushes probably looks like complete mulch. I'm not surprised light.exe has no idea what to do with it.

take metlslime's suggestion, make the whole thing a func wall. and put skip brushes behind it so it doesn't leak, of course. 

This is old now but stands true on most cases:

http://user.tninet.se/~xir870k/tooltips.txt

It should you understand whatever errors you're getting. 

Processing hull 1...
*** WARNING 09: Couldn't create brush faces
*** WARNING 09: Couldn't create brush faces
Processing hull 2...
*** WARNING 09: Couldn't create brush faces
*** WARNING 09: Couldn't create brush faces

I started getting these after removing a brush that wasn't closed. 

* Use different compilers, see if the problem disappears
* Contact compiler author if there's no obvious remedy - maybe it's a tool problem or can be fixed through a code change.

@ModFax :
Is the eMail address in your func_user information still accurate ? 

http://tomeofpreach.wordpress.com/2014/06/26/unusual-func_door_secret-angles/

That's how you do a func_door_secret that moves up last. Hardly even feels like a map hack, but there it is! 

Neat! 

To quote a household name, thks Preach ! 

@mtlslm- turned it into a func_wall but nothing changed.
I thought func_wall didn't lit light.

@rebb- I can do, but I was wondering why the compiler give normal results, before the error started.
As long as you don't send me mod faxes I'm reliable on that email adress. 

that looks similar to a bug I was helping mfx with: https://sourceforge.net/p/quakespasm/patches/15/attachment/mfxsp17_before.jpg - In that case it was an engine bug, but I think it could also be caused by the light tool being compiled with sse2. which light tool are you using? 

@MadFox, if you'd be willing to send me the map source and compile tools you're using, i'd be interested to poke around and see if I can see what's happening