Subject: [Gtkradiant] [Bug 583] Drag edges corrupts brush while changing grid size and manipulating

[jeremiah sypult]

Ahoy,

> Reason this happens is because of radiant's snaptogrid policies.
> if you move an edge, it will try hardest to snap the next closest
clockwise edge...
> if i remember right it's clockwise.

> anyway, it surprises me that when you change the grid size from 4 to 16
that it didn't happen then.
> as you're moving one end about, the other end snapstogrid 2 edges that
wind up on top of each other.
> And since radiant doesn't decimate verts or edges when they're put like
this, the brush blows up.
> You'll likely have the same thing happen in vert-edit.

*nods*

Indeed.

When I started noticing it was when I was doing some editing on some simple
rectangular glass-type brushes. They were 2 units thick, about 128 units
wide, and between 32 or 64 tall. I started noticing it at times when i'd
switch from 1 to 2 to 4 to 8 grid sizes... In my specific case it was
happening fairly regularly with a specific brush never exceeding an 8-unit
grid.

What was also strange was I could move to another identical brush brush
(rotated 90 degrees on an L shape) and do this without any problems. If I
went back to a specific brush it would 'blow up'. This would happen even if
I abandoned my changes and reloaded the map. It seemed specific brushes had
problems with edge-handling while others were seemingly fine. I've even
played with the bobtoolz brush-cleanup to see if it was a strange invisible
brush.

When I saw the erratic behaviour I started messing with radiant to try and
recreate the problem as I've never encountered it before... The way I
explained was the quickest way to demonstrate the problem.

> the programming solution to your problem isn't just fixing YOUR problem,
> it's changing a shit-ton of snaptogrid code to act conditionally.
> some people wouldn't mind snaptogrid with decimation,
> some would want points offgrid to stay offgrid,
> and some would want it to snap AWAY from other verts/edges...

Understood. Is there some type of method that one could use to work around
this problem? Radiants stability seems to be in question once this
strangeness starts to happen...

Thanks much for the reply,

.jer