Pages: << Previous 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 Next >>
Summary Period: 2000-05-16 to 2000-05-11 (Commits 2495-2544 of 12744)
-*- Eliminated all the remaining compilation errors and warnings in the
new terrain library. This mostly involved missing I/O streams, lack
of bool type (on some platforms), implicit conversion from float to
integer, etc.
-*- The terrain makefile now defines __CRYSTAL_SPACE__ when building for
Crystal Space. Currently this define is used to control whether or
not cssysdef.h and cstypes.h are included.
-*- Enabled all the code in csengine/terrain.cpp which was protected
within DDG_FIXME blocks.
-*- Updated makefile so that Crystal Space applications once again link
against the terrain library.
1 lines of code changed in:
Terrain updates by Eric and Jorrit.
14 lines of code changed in:
Update.
3 lines of code changed in:
- Fixed csPolygonTree::ClassifyPoint().
- Disabled the new routine to calculate the solid mask because
it is too slow. I need to make this faster.
37 lines of code changed in:
- Added csOctree::ClassifyRectangle() that will be used for
PVS calculation. This function calculates a 4x4 bitmap representing
the solid space in some rectangle.
- At octree calculation time the solid masks are precalculated
for every node. The octree cache format version has been bumped
up so that recalculation of the octree is automatic.
- Modified db_octree command so that it will show all leaves
(disregarding depth) if -1 is given as an argument.
30 lines of code changed in:
Added csPolygon3D::IntersectRayNoBackFace() which does the
same as csPolygon3D::IntersectRay() but without back-face
culling.
11 lines of code changed in:
Small quality increase.
1 lines of code changed in:
Removed a few warnings in terrain.cpp.
11 lines of code changed in:
Continue integration, files should compile cleanly even with DDG_FIXME enabled
76 lines of code changed in:
Minor changes/cleanup
2 lines of code changed in:
Update.
1 lines of code changed in:
Renamed the following files and fixed references to them throughout the
project:
def.h --> csdef.h
sysdef.h --> cssysdef.h
types.h --> cstypes.h
frustrum.h --> frustum.h
frustrum.cpp --> frustum.cpp
1469 lines of code changed in:
Implemented new 'TEXTURE' keyword for upcoming terrain engine.
With this keyword you set the basename for all the textures
used by the terrain. The base should be something like
'MyTexture%d.jpg'. %d will be replaced with 0 to 127 for example.
23 lines of code changed in:
Add stubs for AddWorld and AddSector
8 lines of code changed in:
- Added const version of csPoly3D::operator[].
- Added csPoly3D::SplitWithPlaneXYZ().
- Implemented csPolygonTree::ClassifyPoint() and csBspTree/
csOctree::ClassifyPolygon. This functions respectively test a
point/polygon against the tree. The test returns wether or not
the object is in solid space.
277 lines of code changed in:
Cobbled things back together just enough so that the project can once
again be built on Unix following phase one of Alex's merge of the new
terrain code. For now terrain functionality is completely disabled.
For Windows, the project file maintainer will probably need to adjust
include paths in the csparser and csengine projects to refer to
libs/csterr.
13 lines of code changed in:
Cobbled things back together just enough so that the project can once
again be built on Unix following phase one of Alex's merge of the new
terrain code. For now terrain functionality is completely disabled.
For Windows, the project file maintainer will probably need to adjust
include paths in the csparser and csengine projects to refer to
libs/csterr.
38 lines of code changed in:
- Fixed bug in PVS. Correctly does backface culling now.
- ProjectAxisPlane() in csgeom now returns a boolean to indicate
if the projection was possible.
- Did some fixes on PVS again. For the very first time PVS seems
to be doing what it is supposed to do: cull invisible nodes :-)
Quality of the PVS is still bad (too few invisible nodes) but
Wouter and me discussed various techniques to solve that.
Note that polygons are not yet included in PVS. This is only
the node -> node PVS part.
- Added PVS optimization suggested by Wouter Wijngaards which
treats nodes that cannot see an occludee as completely solid.
This both makes PVS creation faster and increases the quality
of the PVS as well. This is still not the full set of quality
improvements that is planned.
- Added csPoly3D::ClassifyXYZ() routines.
- Started work on new optimization for PVS.
287 lines of code changed in:
Fixed bug in OpenGL makefile. The inclusion of the software
renderer caused problems if the CS version of NASM was present.
The NASM asm files were not included in the OpenGL makefile
causing undefined symbols in the included software renderer.
19 lines of code changed in:
Added Lock functions to skeleton
4 lines of code changed in:
Disabled the terrain engine in preparation for the move over to the new version.
235 lines of code changed in:
Repaired several problems in the software renderer which have been
present since Andrew's most recent overhaul.
-*- Fixed rendering of fog in 32-bit mode on NeXT for both big- and
little-endian. There were several problems:
+ Inconsistent interpretation of TOP8BITS_R8G8B8_USED. When
Andrew first introduced TOP8BITS_R8G8B8_USED it was interpreted
as meaning that the display driver might _possibly_ use the high
byte of a 4-byte pixel value. However, in some of his more
recent code, it was interpreted as meaning that the high byte
was _positively_ used. As a consequence, DrawFogPolygon() chose
completely incorrect values for csScanSetup's FogR, FogG, and
FogB values on little-endian machines.
+ Inconsistent interpretation of csScanSetup's FogR, FogG, and
FogB values. In some cases, the fog scanline functions expected
these values to be normalized to 0x00RRGGBB (note that the high
byte is unused), but DrawFogPolygon() initialized these values
without normalizing them. Furthermore, even the scanline
functions were inconsistent in their expectation of these
values. In particular, the calculation of 'fog_pix' neglected
to take normalization into account.
+ Incorrect definition of PIXEL_PREPROC() and PIXEL_POSTPROC().
For instance, PIXEL_POSTPROC() was defined as "x << 8". When
invoked with PIXEL_POSTPROC(r|g|b) this had the unfortunate
side-effect of only shifting the blue color rather than the
composite value, since it expanded to the equivalent of
(r|g|(b<<8)). The correct definition should have been
"(x) << 8", which would expand to ((r|g|b)<<8) as desired.
+ Signed rather than unsigned values in several places, including
the scanline routines as well as csScanSetup's RGB values.
During right-shift operations, this could result in unwanted
sign-extension.
-*- Fixed a bug in the fog scanline drawing routines where they failed
to clamp the blue color to 0x000000ff when recomposing the
'destination' pixel. (Note that the red and green colors were
correctly clamped to 0x00ff0000 and 0x0000ff00, respectively.)
-*- Fixed rendering of halos in 32-bit mode on NeXT for big-endian
machines. There were several problems:
+ Overflow during computation of a halo color component, which led
to corrupt composite color. The problem was that
csSoftHalo::Draw() was computing color values based upon color
masks using the formula "(int)(C * pfmt.mask + 0.5)" where C is
a floating point number. On big-endian NeXT machines, where the
red mask is 0xff000000 this formula overflows (even in the
obvious case when C is 1.0).
+ Inconsistent interpretation of csScanSetup's FogR, FogG, and
FogB values. The halo scanline functions expected these values
to be normalized to 0x00RRGGBB (note that the high byte is
unused), but csSoftHalo::Draw() initialized these values without
normalizing them.
+ Signed rather than unsigned values in several places, which
could have resulted in unwanted sign-extension.
-*- Fixed trilinear filtering for 32-bit big- and little-endian on
NeXT. The problem was that the 32-bit 'filt2' routines neglected
to apply PIXEL_POSTPROC() to the composed pixel value.
235 lines of code changed in:
Reverted Paul's most recent change for a couple reasons:
1) He forgot to commit all the other files upon which this change depends.
2) He accidentally blasted all the useful default values which are
suitable for most people and replaced them with his custom settings.
21 lines of code changed in:
Reverted Paul's accidental commit of cryst.cfg.
3 lines of code changed in:
Mingw/NT4/Win98 Update
-- user.mak commented to ignore PLUGINS reference. PLUGINS are
determined by Mingw.mak, win32vc.mak, unix.mak, etc. If you
require that PLUGINS be referenced in user.mak, then
uncomment PLUGINS references in user.mak, etc.
42 lines of code changed in:
Reverted Paul's change since it wasn't necessary. mk/cs.mak automatically
disables MAKE_DLL when appropriate (such as when USE_SHARED_PLUGINS is
disabled), so there is no need to do so manually here.
0 lines of code changed in:
wouter accidently renamed updateclipper to draw
7 lines of code changed in:
Modified to once again allow for all required source files
in plugins/video/canvas/common
Extended vpath to include plugins/video/canvas/common
4 lines of code changed in:
Updated Mingw
7 lines of code changed in:
Fixed file truncation
-frank (frank@oconnors.org)
1 lines of code changed in:
Warning level is now configurable. Failure to change directory group
ownership and permissions no longer generates warnings at default level.
This was changed because SourceForge now runs a daily job which assigns
ownerhip of all files and directories in the FTP area to owner "ftp".
Thus, once "ftp" owns the directories, the chgrp and chown commands which
this script uses will fail.
14 lines of code changed in:
Updated to accomodate static libraries for Win32/GCC
6 lines of code changed in:
Updated to accomodate DD_FALSE for OS_WIN32 platforms
which do not have support for certain versions of DirectX
3 lines of code changed in:
dangling else fixed (added {}).
3 lines of code changed in:
Cleaned up the code a lot, and removed various unused routines.
This should be faster as well.
Ambient is calculated better (better formulas)
as is the priority.
Did the fix as discussed with AndyZ.
Downsides:
- still in 64x64 on flarge, some array bounds errors occur.
(And I provide the correct frustum now...)
- And there seems to be a blue-attack in room and flarge.
Either an error in the code, or an intensely blue polygon somewhere...
Perhaps the sky polygon should be higher in the sky - if this now has a
lightmap - distance will fade the light.
(It seems to be looping on a blue polygon, which sheds light, then gets it
reflected back, and reflects that light ... etc.)
112 lines of code changed in:
Added SetLastFrustumId(int) public routine.
3 lines of code changed in:
A bunch of BeOS fixes and improvements by Eric Sunshine.
11 lines of code changed in:
It appears that Jorrit accidentally committed this file.
2 lines of code changed in:
Removed CHK and CHKB.
903 lines of code changed in:
Moved declaration and definition of cspExtDialog outside of the function
within which it was defined since the BeOS compiler was generating
unlinkable code.
52 lines of code changed in:
Eliminated compilation warnings.
1 lines of code changed in:
-*- New CSWS component: notebook control AKA "property shit".
Only used in cswstest for now. Going to be used in MazeD.
1477 lines of code changed in:
More Eric stuff.
7 lines of code changed in:
Update.
2 lines of code changed in:
- Worked a bit more on PVS. Still not working :-)
- Made PVS caching more robust.
17 lines of code changed in:
Fixed bug: Push() wasn't abstract.
9 lines of code changed in:
- Added 'freezepvs' command to freeze the PVS at some position. This
is useful to debug the PVS. After freezing the PVS you can walk
around and examine what is visible and what not.
- Added 'pvsonly' command so that you can force visibility culling
to operate ONLY with the PVS and not use any other culling
mechanisms (like c-buffer).
150 lines of code changed in:
Fixed a bug introduced by Paul
1 lines of code changed in:
added SetFontSize
9 lines of code changed in:
components can now draw with their own fontsize - added SetFontSize ()
40 lines of code changed in:
Pages: << Previous 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 Next >>
Generated by StatCvs v0.2-dev