Generated by StatCvs v0.2-dev
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: 2000-06-02 to 2000-05-28 (Commits 2645-2694 of 12744)
-*- Fixed the CS_ASSERT() macro so that it now works as expected; that is,
it "asserts" that some condition is true. Previously, the logical
sense of the conditional test was reversed.
-*- Enhanced CS_ASSERT() so that it now prints a useful error message
before breaking into the debugger or crashing the application, instead
of dying silently.
16 lines of code changed in:
A few small modifications to PVS building and some experiments.
38 lines of code changed in:
Some changes which gets the metaballs texture working. Mainly to do with many iGraphics3D refering to one iGraphics2D
-The engine now keys of iGraphics3D rather than iGraphics2D, to determine whether it needs to swap states.
-Metaball plugin now queries iGraphics3D for width and height of context.
-Opengl: need to convert vertices co-ordinates to the upper left subregion of the backbuffer
97 lines of code changed in:
Workaround to allow Mac users to see text; apparently the current Mac drivers
do not support a pure GL_ALPHA texture, so for the OS_MACOS platform we
use GL_LUMENENCE textures instead. This means that text is not shaped (all
letters are on a black background instead of showing the picture behind
them) but at least you can see the text!
22 lines of code changed in:
Now dynamic textures clean up after themselves properly.
37 lines of code changed in:
Fixed the 'clean' makefile target for Simple.
1 lines of code changed in:
Fixed a nasty bug where VFS was being created and destroyed twice in
applications which requested VFS explicitly with RequestPlugin(). In
addition to other problems caused by double instantiation, this had the
particularly undesirable effect of creating and destroying VFS's global
ArchiveCache twice, thus resulting in a crash on some platforms.
19 lines of code changed in:
Worked around a NextStep compiler shortcoming where it failed to grok
subclasses of csFrustumView::CleanupAction.
30 lines of code changed in:
Eliminated compilation warning about unused variable.
0 lines of code changed in:
Fix for software rendering.
27 lines of code changed in:
Renabled fixed frustrum clipping for crystal space, fixed rendering.
32 lines of code changed in:
Various fixes, migrated the behave well under cantankerous window managers code into glx
67 lines of code changed in:
Reverted Andrew's changes to csVector::Find() and FindKey() since they
broke other parts of the system. He changed the functions so that they
search the vector from tail to head rather than head to tail.
Unfortunately, other parts of the system (even Andrew's own code) rely on
the head to tail order. An example of this is csSystemDriver's
GetOptionCL(), which broke after Andrew's change.
7 lines of code changed in:
Reverted Andrew's changes to Find() and FindKey() since they broke other
parts of the system. He changed the functions so that they search the
vector from tail to head rather than head to tail. Unfortunately, other
parts of the system (even Andrew's own code) rely on the head to tail
order. An example of this is csSystemDriver::GetOptionCL(), which broke
after Andrew's change.
2 lines of code changed in:
Fixed the 'clean' target for ndrvdem.
1 lines of code changed in:
Eliminated compilation warning about uninitialized variable.
1 lines of code changed in:
Fixed out-of-date key names in several configuration files (in data/config)
following renaming which Andrew performed in the source code.
3 lines of code changed in:
Fixed software renderer halos (again) for big-endian 32-bit NeXT,
following Andrew's most recent change which broke them.
14 lines of code changed in:
fixed a msvc5-specific compilation error
14 lines of code changed in:
Eliminated more linkage problems caused by inconsistent use of 'inline'
within csterrain library.
2 lines of code changed in:
Eliminated compilation warning in polytext.cpp about unused variable.
Worked around bug in OpenStep compiler triggered by polytext.cpp where
unlinkable code is generated when an inline method calls a second inline
method before the compiler has seen the second method. Textually swapping
the methods fixes the problem.
3 lines of code changed in:
Fixed compilation errors on some platforms in soft_txt.cpp where it was
trying to implicitly cast void* to RGBPixel*.
30 lines of code changed in:
- Added csBox2 and csBox3::Contains() functions to test if some box
contains another one.
- Fixed PVS test with adjacent nodes. It was not correctly testing
if one box contained the other.
48 lines of code changed in:
-*- Removed the unneeded VFS->IncRef in systemdriver::initialize.
This removes the "1 unreleased instances" error at cleanup.
-*- Fixed the vshell plugin (it was creating a second instance of VFS
although VFS plugin is written such that it allows just one instance).
-*- Slightly improved csFlags class. WARNING: The former Set(int) method
is renamed to SetAll(int) now, Set(int) has a different meaning!
-*- Moved bool csPolygon3D::dont_draw and bool csPolygon3D::delete_portal
to csPolygon3D::flags. Another 8 byte win for csPolygon3D.
-*- Added macro CS_ASSERT for system-independent assertions. It will fire
a debug breakpoint if the assertion fails (rather than printing a error
message which is useless in CS for many platforms).
-*- There is also a DEBUG_BREAK macro (which is empty for non-debug mode).
You can use it if you want to fire a debug breakpoint under some
complex conditions (for simple conditions you may use CS_ASSERT).
Currently it is defined only for x86 processor.
-*- Found and fixed a bug (that maybe the one reported by Jorrit in past)
with VFS creating invalid archives. The problem was that the contents
of "lazy" array were simply appended to "dir" array (after
csArchive::Flush) while they should be InsertSort'ed.
-*- Removed csFrustum::Intersect and csFrustum::IntersectFull methods.
Added instead csFrustum::Classify() which is faster and less buggy.
There is still a bug somewhere but its already better than the old one.
-*- Caught and fixed a bug in software halo drawer introduced by Eric.
-*- Many other small changes I don't remember now.
722 lines of code changed in:
- For debugging: control precisely what parts of PVS are done
and what not.
- Added 'bool cd' parameter to csCamera::Move and MoveWorld to
allow moving around the world without collision detection to
sector walls and vice versa. Portals are still checked though.
- Fixed collision detection option in WalkTest so that it now
also works for STATBSP levels (i.e. you can walk through
things).
- Fixed db_octree console command in WalkTest so that you can
actually use db_octree -1 to show the leaves.
- Fixed a serious bug in PVS again. The solid boundary optimization
was in fact used completely wrong. Unfortunatelly this decreases
PVS quality a bit. PVS is still not completely right though.
- Changed PVS building so that first pass is now the solid space
PVS pass and second pass is full PVS. This improves speed and
quality a bit.
- Changed solid space point tester so that it now requires a
number of directions for which the solid space test succeeds.
This decreases the chance of an error cause by bad polygons
in the input data.
- Extended dump of octree nodes with the values of the calculated
solid masks for the six directions.
- Added a new allegro canvas driver from Burton Radonsxi
<loth@pacificcoast.net>. This allegro driver is developed on Linux
and should work on Linux. It will probably not replace dosalleg
yet since the new one depends on iunix.h. We're looking into that.
99 lines of code changed in:
Revisions to accomodate auto-gen of include\volatile.h
for Mingw.mak
3 lines of code changed in:
added some comments
4 lines of code changed in:
Worked around NextStep compiler limitation with regards to multiple-
inheritance. It was crashing on bare call to SetPosition() which is
declared in both iParticle and csSprite. Now makes explicit call.
19 lines of code changed in:
Now that solid space calculation works I fixed PVS to use it
in a safe way. Solid space for node boundaries is not reflexive.
Even if node A cannot see B it is still possible that B can see A.
3 lines of code changed in:
Undo.
2 lines of code changed in:
- Fixed terrain engine so that it renders triangles again. Still
have to work a bit on speed and other issues.
- 'Probably' fixed the solid space calculation for real now :-)
- Removed files from apps/metademo which are now in the plugin.
11 lines of code changed in:
Got a large patch from Samuel Humphreys with the following changes:
- Adds a couple of new walls, bots and another dynamic texture
to dtsimple. The new dynamic texture is rendered by the engine
as the bots point of view, or as a manually configurable view
from (0,0,0). Only works with the software drivers currently
in 8, 16 & 32 bit modes. The opengl software texture version
will not work...it will crash. The backbuffer opengl version
will not work, but won't crash. I think the glide version will
work, can't test this though.
- Added new flag to iTextureManager: CS_DYNAMIC_TEXTURE_SHARE_HINT.
See explanation there.
- Added many more keys to fiddle with in dtsimple.
- Added another software dynamic texture implementation to the
software drivers. This implementation shares the texture manager
with the parent context. This allows you to render to it with
textures that you have already registered with the parent
context i.e. a map view courtesy of the engine.
- In pursuit of this added functions to the software 3d drivers
which allows the initialisation and opening of the driver
according to another driver while sharing resources. The
sharing software dynamic texture utilises these functions.
This is also in preparation for the window manager...coming
soon!
- Pluggified metaballs and updated metademo and dtsimple as
appropriate.
- Added more parameters to vary to see what happens in metademo.
- Added a couple more functions to iSystem, which are implemented
in system.h/cpp, so shouldnt break anything. These are queuing
functions for two new windows messages.
- Added two new windows messages. ..ContextResize and
..ContextClose.
- Updated the engine to deal with new messages
- 'Multiplexed' the engine. Added new function
csWorld->SetContext(iGraphics3D *g3d). This allows the engine
to swap states according to g3d. The new dynamic texture
demonstrates this.
- Fixed the annoying bug in the Xlib software drivers where on a
fresh Xserver session the window placement wouldn't behave for
the first three or four times. Borrowed the trick from gdk :).
Much more window manager friendly now...i.e. Crystal Space is
beginnning to behave politely!
- Migrated XLib changes into the line drivers.
- Added resizing to the line drivers.
- Got rid of a static file scope variable in the software
texture manager.
- Small makefile fixes for blocks and csauth.
- Did a fix to dump(csPolygon3D*) for polygons with no name.
779 lines of code changed in:
More terrain fixes.
7 lines of code changed in:
- Found a serious bug in PVS (more exactly the calculation of solid
space between nodes). Note that I didn't fix the bug yet :-)
- Added more debugging information when dumping the octree/bsp tree.
- Added a number of 'const' declarations to members in csPoly3D.
- Extended BSP tree traversal with an extra flag 'same_plane' so
that the BSP traversal routines know if all given polygons are
on the same plane or not. This could be used for optimizations.
- Changed BSP building so that it better detects the case where
all polygons are coplanar and indicates that in the node.
- Added csPolygon3D::IntersectRayPlane() to intersect a ray
with the plane of a polygon.
77 lines of code changed in:
More fixes by Eric.
10 lines of code changed in:
Committing a debug version, so I can take it home from cvs :)
2 lines of code changed in:
- Fixed a number of warnings in the WalkTest changes from David
Durant.
- Still show the warning when RAPID plugin cannot be loaded by
WalkTest but doesn't fail anymore.
- Fixed an inline problem in DDG terrain engine and a small other
fix.
17 lines of code changed in:
Minor fixes/cleanup.
86 lines of code changed in:
Redoing myapp functionality.
16 lines of code changed in:
Redoing myapp functionality.
0 lines of code changed in:
Small fix to terrain engine (removal of bogus 'inline').
23 lines of code changed in:
Enable all terrain near the viewer.
2 lines of code changed in:
Eliminated compilation warning from egcs-1.1.2.
Added a lot of missing 'const' qualifiers.
35 lines of code changed in:
Added another myapp callback.
4 lines of code changed in:
Even more "myapp hack" fixes.
13 lines of code changed in:
Eliminated a slew of compilation problems.
David Durant, please remove this "myapp" hack. This is not the proper way
to provide extensibility. Use subclassing instead.
42 lines of code changed in:
Added myapp hook functionality.
69 lines of code changed in:
- Created new 'perftest' application to test the performance
of a 3D renderer. This application will try various tests to
see how fast the 3D renderer is. 'perftest' is not finished.
Only a small fraction of the 3D capabilities are tested.
- Changed the way csPolyTree::ClassifyPoint() works. Instead
of assuming space is open and testing six directions to prove
it is solid (all six directions must be solid) we now
assume space is solid if we find only one direction for
which space is solid.
- Added csPoly3D::GetCenter() which computes the center of the
polygon and returns it.
- Improved the solid space generator so that it is now much more
likely to say that a node boundary is solid. This should improve
PVS considerably. Unfortunatelly it also seems to contain a few
bugs so at the moment it is not useful yet.
607 lines of code changed in:
Stub functions for the app developer.
20 lines of code changed in:
This file is #included in walktest.h
within the definition of the class
WalkTest. The app developer may add
members to WalkTest here.
10 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