seanisom / vtp Goto Github PK

What steps will reproduce the problem?

1. Compile "Simple" project in debug mode and its dependencies (vtdata, vtlib, 
xmlhelper)
2. Launch "Simple.exe"

What is the expected output? What do you see instead?

The program should launch and create the terrain according to the "Simple.xml" 
configuration file. Instead of that, the program segfaults (see attachment). 
After debugging, it appears the problem is at line 1043 of the 
vtlib\vtosg\Mesh.cpp file, vtMesh::AddStrip method :

1038:   osg::UIntArray *uia = getIndices();
1039:
1040:   for (int i = 0; i < iNVerts; i++)
1041:       uia->push_back(pIndices[i]);
1042:
1043:   getDrawArrayLengths()->push_back(iNVerts);  <- this line

I could see that getDrawArrayLengths() returns a non-null pointer. It fails 
during the push_back (I've checked by initializing a local variable and then 
push_back).

What version of the product are you using? On what operating system?

vtp-src-110619.zip
mini 9.8, built from sources
OpenSceneGraph 2.8.5, built from sources
GDAL 1.8, built from sources
WinAPIs-1104.zip
Windows XP Pro SP2 32 bits

Please provide any additional information below.

1. Launch enviro with a terrain which has a derived texture.
2. Turn on the Elevation Legend.
3. Switch to another terrain.

In fact, not only does the legend not update for another terrain, it
doesn't even hide itself if you go to Earth View.

Seems a findGLUT.cmake is needed.
When compiling in windows with GLUT, glutSimple would error at linking, not 
being able to find a couple of calls in a library. Tracked that library down to 
being glut32.lib.
Looking at the CMakelists.txt for glutSimple, there's a clause around all the 
library includes "if(GLUT_FOUND)". Searching for GLUT_FOUND reveals it's not 
set on anywhere.
I've temporarily added a bool entry to CMAKE to handle this. 

Would love to help here, but I have no idea how. Above my knowledge I'm afraid!

I am using the opensuse build service to produce vtp rpms, and the build
service found an issue with file reading. 

What steps will reproduce the problem?
1. compile rpm on opensuse build service (home:repabuild project on
build.opensuse.org)
2. inspect results for opensuse 11.1 build

What is the expected output? What do you see instead?

Compiled rpms with vterrain

I: Statement might potentially overflow a destination buffer, where a size
larger
than the actual buffer was specified
E: vtp destbufferoverflow ElevationGridIO.cpp:1221, 1222
E: vtp destbufferoverflow RoadMap.cpp:766, 868

line 868 is:
            fread(&stmp, shortSize, 1, fp);             
RoadMap.h:#define shortSize 4

and stmp is declared as:
    short stmp;

Yes, it is amazing how critical the rpm build process for opensuse 11.1 is,
but it has a point, since this code tries to load 4 bytes in a 2-byte
variable.  

What version of the product are you using? On what operating system?

vtp-081224

opensuse 11.1

Please provide any additional information below.

What is correct, what is wrong. I guess that for the file format, shortSize
should be 4? In that case, i suggest using (at least on linuxes etc)
inttypes.h, and int32_t etc. to really fix the variable size.

What steps will reproduce the problem?
1. Launch Enviro
2. Use mouse to advance forward. Maintain mouse position.

What is the expected output? What do you see instead?
A continuous updating canvas; however it freezes until there is some other
event - a mouse movement, key input, etc.

Possible solution? 
Include a timer/update loop.

What steps will reproduce the problem?
1. In VTB, load any elevation layer with undefined areas (such as most any
USGS DEM)
2. Load any vector layer on top of it
3. Zoom in well past 1:1.
4. Notice that the elevation layer is drawn badly offset (relative to the
vector layer which is drawn correctly.)  The offset gets worse the further
you zoom in.

Apparently it is due to some integer coordinate rounding error in the 2D
graphics pipeline, either in GDI or in the way wxWidgets calls GDI.

It appears to be a basic Windows limitation: GDI won't stretch and mask at
the same time, which forces us to use wx's SetUserScale, which call's GDI's
SetViewportExtEx, SetWindowExtEx, SetViewportOrgEx, SetWindowOrgEx.  The
roundoff problem seems (?) to be related to those methods.

For some reason the multi-threaded updateroi call is missing most of it's 
arguments.

1. Load any elevation layer.
2. Menu command: Elevation: Render to Bitmap.
3. The result is a solid black bitmap, instead of the expected color data.

BobMaX Test to add DAE. WRL, OBJ export

What steps will reproduce the problem?
1. From menu Guide -> Documentation (Local)
2.
3.

What is the expected output? What do you see instead?
The expectes output is the visualization of the guide. A error (no file 
found) is show.

What version of the product are you using? On what operating system?
Enviro . jan 13 2008 on Windows XP

Please provide any additional information below.
I think that the error is in the path of the guide in enviro code. If I 
resolve the path from myself it's work.

Best

Inserted DAE and OBJ extensions

What steps will reproduce the problem?
1. Problem found on code instpection of Contours.cpp

What is the expected output? What do you see instead?
I do not believe this is correctly looping through all tiles.

What version of the product are you using? On what operating system?
Contours.cpp - r7558

Please provide any additional information below.
For loop on Line#59 when calculating the minLod
"tiledGeom->rows * tiledGeom->rows"

Should this be "tiledGeom->cols * tiledGeom->rows"

MiniDatabif.h has:  m_min = new int [cols*rows];

Possible issue if cols is smaller then rows
Example: cols = 3 rows = 4
loop will access array 16 times when only 12 items are available

What steps will reproduce the problem?
1. Create a terrain with a projected like IBCAO data (around the pole).
2. Launch Enviro in earth view.
3. See the red rectangle for the terrain, around the pole.
4. Put the cursor inside the rectangle.
5. Notice that the terrain is not detected.

The problem is we do point-in-rectangle test in lat-lon coordinates.  But,
some areas (especially around the poles) do not map to lat-lon rectangles.

The solution is implementing an inside-test for true spherical polygons. 

Added VTLOG  XYZ coordinates  on click

What steps will reproduce the problem?
1. Launch VTBuilder
2. Load data
3. Try to draw a rectangle with the area tool.

What is the expected output? 
A rectangle that follows the mouse.

What do you see instead?
A vertically down shifted rectangle.

What steps will reproduce the problem?
1. In Global Options, turn on 'Use direct picking of object under 2D cursor'.
2. Launch a terrain.
3. Create two instances of the same 3D structure model.
4. Try to pick them to move them.  Note that one of the objects isn't pickable.

This is just one of the cases where 'direct picking' is not consistent or
feature-complete.  RJ wrote it, but it was not finished to handle all
possible cases of selection.  Since then, we recommend users to simply not
use it.  We should eventually either fix this capability, or remove it.

Inserted definition for WriteDAE, WriteWRL and WriteOBJ from TIN

What steps will reproduce the problem?
1. Run CManager.
2. Drop in a .3ds file (I used "ITO.3ds", which was exported from 3DS MAX.)
3. Toggle wireframe on and off until it crashes in OSG's State management.

What steps will reproduce the problem?
1. Open the feature table dialog for any Raw or Abstract layer that has
several fields.
2. Scroll the dialog to the right by any amount.
3. Right-click a field to edit its value.  Note that that the wrong field
can appear, because the code is using the unscrolled rather than the
scrolled mouse X value to determine field.

1. Load any road network (RMF)
2. Select any node, right-click for Node Properties.
3. The 'View' pane should show the roads meeting at the node, but it is way
off, it has to be panned w-a-y around to eventually find it.

Might be an issue with wxWidgets changing how scrolling of embedded panes
works.

"Color map file" combo box from "View Options"/"Rendering options..." dialog is 
empty.

In VTBuilder Elevation/Export To... PNG gives image with wrong looking crests 
and peaks.

Inserted WriteDAE, WriteWRL and WriteOBJ   from TIN

Inserted DAE, WRL and OBJ   Export from TIN menus

I'm trying to create vterrain package in Arch Linux distribution. I have 
created some of the dependencies:

LibMini https://aur.archlinux.org/packages/libmini-svn/
Libgrid https://aur.archlinux.org/packages/libgrid-svn/

The other dependencies are the following:

OpenSceneGraph 3.2.1
wxGTK 3.0.2
GDAL 1.11.1
proj 4.8.0-3

However, compiling reports some errors that I could not to identify. The build 
log is below:

https://gist.github.com/samtux/46809c4e6828fef362bd

Any help is appreciated. I would like to create the package for Arch Linux 
users.

regards,

Samuel Mesa.