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: 2000-05-11 to 2000-05-09 (Commits 2445-2494 of 12744)
components can now draw with their own fontsize
20 lines of code changed in:
Removed "ifdef" to allow all compilers to use
new virtual path reference.
0 lines of code changed in:
added method to set fontsize
4 lines of code changed in:
more robust.
3 lines of code changed in:
Now, the bug is almost fixed. And I know why it happens.
The light frustum given to GetCoverage must be clipped to
responsability rect, and this was not done yet.
csPolyTexture::GetLightmapBounds will need some small
change to work with this (the frustum_id check / code
must go back to csPolygon I think.)
13 lines of code changed in:
specified new location of glide renderer
1 lines of code changed in:
Added 'pvs' command to toggle pvs usage (note that PVS is still
not working properly).
20 lines of code changed in:
Changed calloc/free in polytext.h (csCoverageMatrix) to
new[]/delete[] in order to help debugging it (memory debugger
only works on C++ memory allocators).
5 lines of code changed in:
Removed some old stuff.
0 lines of code changed in:
Update.
3 lines of code changed in:
fixed delete to delete[].
1 lines of code changed in:
Added quick and dirty memory debugger in WalkTest that's a bit
easier to setup than the one from Andrew and also works immediatelly
in a debugger. Just go to the end of WalkTest and change #if 1
to #if 0. It catches memory overwrites, double deletes, and
allocs with 0 bytes.
62 lines of code changed in:
Finished caching of PVS. Also added dummy PVS calculator that
is used for now and just creates a PVS that says that everything
is visible :-)
83 lines of code changed in:
Somewhat better body proportions for flarge
3 lines of code changed in:
Made player's body somewhat smaller, now you can even jump (not only pass
through the corridor) in mazedex
8 lines of code changed in:
Modified to accomodate for GCC only.
-- changed vpath to consider plugins/video/renderer/common
as well as plugins/video/renderer/software
11 lines of code changed in:
added a check before calling DecRef in destructor
-frank (frank@oconnors.org)
2 lines of code changed in:
Works with dynamic textures now.
-frank (frank@oconnors.org)
12 lines of code changed in:
Re-added the explicit IncRef() on VFS (at least for now). After Norman
removed it, command-line applications began crashing at termination time
since they were sending an extra DecRef(). For now, I prefer to have the
application leak the VFS object rather than actually crash. We need to
resolve this reference count issue once and for all, though. Since command
line programs crash, but graphic programs don't, I might suspect that one
of the plug-in modules (which command-line programs don't use) is leaking a
reference to VFS, though a scan through them reveals no obvious errors.
2 lines of code changed in:
Quick fix for BeOS.
2 lines of code changed in:
-*- Overhauled the approach used for dynamic (plug-in) linking on MacOS/X
Server, OpenStep, and NextStep ports of Crystal Space. These ports
now do a fairly good job of emulating dynamic linking as it is done on
most other platforms in which symbols within a plug-in are private
unless explicitly exported. Historically, with the NeXT ports, all
symbols were automatically public. This naturally leads to severe
symbolic conflicts since many plug-ins link with the same static
libraries. As a consequence, in the past I had to devise elaborate
mechanisms to work around these conflicts.
-*- The NextStep port emulates private plug-in symbols by actually
stripping all public (resolved) symbols, except for the SCF
GetClassTable() hook, from the plug-in module at build time. This
technique provides fairly decent insurance against symbolic conflicts
at plug-in load time even when multiple plug-ins have linked against
the same static libraries. The only big drawback to this approach is
that it makes symbolic debugging impossible. I had tried the
alternate approach of parsing the error stream from RLD and calling
rld_forget_symbol() for symbolic conflicts, and then attempting a
reload, but the RLD run-time is sufficiently buggy that it would
actually crash within rld_forget_symbol().
-*- The MacOS/X Server and OpenStep ports emulate private plug-in symbols
by taking advantage of DYLD's NSLinkEditErrorHandlers which allows
the loader to ignore duplicate symbols at load time, thus there is no
need to strip them away at build time. This approach has the
advantage that symbolic debugging capability is retained. I had
attempted to use the same approach as I used for the NextStep port
(that of stripping aways symbols at build time), but discovered that
the Objective-C run-time is more picky in OpenStep than NextStep. In
particular, it couldn't deal with the fact that symbolic references
to Objective-C classes defined locally in the plug-in had been
stripped away. It would be possible to manually instruct the 'strip'
utility to retain certain symbols, but doing so would have introduced
a potential maintanence headache as future changes are made to the
code.
-*- Eliminated all the special-case hacks which had been placed in
various applications to force symbolic references in the executable
itself so that the symbols would be available for dynamically loaded
plug-in modules. Since plug-in modules now link with all the
libraries which they, themselves, require, the application no longer
needs to do so.
-*- Eliminated special-case makefile hacks which forced applications to
link with extra libraries required by plug-in modules which they might
end up loading. In particular, scfreg makefile was considerably
simplified.
47 lines of code changed in:
-*- Overhauled the approach used for dynamic (plug-in) linking on MacOS/X
Server, OpenStep, and NextStep ports of Crystal Space. These ports
now do a fairly good job of emulating dynamic linking as it is done on
most other platforms in which symbols within a plug-in are private
unless explicitly exported. Historically, with the NeXT ports, all
symbols were automatically public. This naturally leads to severe
symbolic conflicts since many plug-ins link with the same static
libraries. As a consequence, in the past I had to devise elaborate
mechanisms to work around these conflicts.
-*- The NextStep port emulates private plug-in symbols by actually
stripping all public (resolved) symbols, except for the SCF
GetClassTable() hook, from the plug-in module at build time. This
technique provides fairly decent insurance against symbolic conflicts
at plug-in load time even when multiple plug-ins have linked against
the same static libraries. The only big drawback to this approach is
that it makes symbolic debugging impossible. I had tried the
alternate approach of parsing the error stream from RLD and calling
rld_forget_symbol() for symbolic conflicts, and then attempting a
reload, but the RLD run-time is sufficiently buggy that it would
actually crash within rld_forget_symbol().
-*- The MacOS/X Server and OpenStep ports emulate private plug-in symbols
by taking advantage of DYLD's NSLinkEditErrorHandlers which allows
the loader to ignore duplicate symbols at load time, thus there is no
need to strip them away at build time. This approach has the
advantage that symbolic debugging capability is retained. I had
attempted to use the same approach as I used for the NextStep port
(that of stripping aways symbols at build time), but discovered that
the Objective-C run-time is more picky in OpenStep than NextStep. In
particular, it couldn't deal with the fact that symbolic references
to Objective-C classes defined locally in the plug-in had been
stripped away. It would be possible to manually instruct the 'strip'
utility to retain certain symbols, but doing so would have introduced
a potential maintanence headache as future changes are made to the
code.
-*- Eliminated all the special-case hacks which had been placed in
various applications to force symbolic references in the executable
itself so that the symbols would be available for dynamically loaded
plug-in modules. Since plug-in modules now link with all the
libraries which they, themselves, require, the application no longer
needs to do so.
-*- Eliminated special-case makefile hacks which forced applications to
link with extra libraries required by plug-in modules which they might
end up loading. In particular, scfreg makefile was considerably
simplified.
2 lines of code changed in:
Update.
2 lines of code changed in:
removed the explicit IncRef of VFS that caused the warning about unreleased instance of kernel.vfs
0 lines of code changed in:
Fixed bug in csFont.
3 lines of code changed in:
Fixed bug: Was incorrectly computing FontCount (one too large). This may
have been cause of crasher in OpenGL AddFont().
9 lines of code changed in:
Fixed printed output.
1 lines of code changed in:
Fixed.
1 lines of code changed in:
Update for Eric's changes.
34 lines of code changed in:
Samuel Humphreys sent me a new patch to clean up his dynamic
textures and also implement them for OpenGL.
173 lines of code changed in:
Update.
2 lines of code changed in:
Restored cs.mak because it isn't needed and isn't helping either.
2 lines of code changed in:
Added dynamic-texture support to NeXT. Removed unused junk from dyntex2d.
7 lines of code changed in:
-*- Coding style fixes for csfont plugin
-*- Miscelaneous fixes to OS/2 DIVE canvas
1113 lines of code changed in:
Fixed an internal compiler error for gcc 2.7.2 in the Crystal
Clear code.
2 lines of code changed in:
Modified makefile system to add an extra 'dyn' or 'stat' in
the 'out' directory to differentiate between dynamic or static
linking.
25 lines of code changed in:
Coding style fixes
42 lines of code changed in:
Eliminated an apparently bogus compilation warning on OpenStep.
1 lines of code changed in:
For dynamic texture system...
-frank (frank@oconnors.org)
12 lines of code changed in:
Neccessary files to compile freefont plugin on win32 with prebuilt
freetype.lib
-frank (frank@oconnors.org)
0 lines of code changed in:
Modified static lib output name to reflect shared lib output name
-- changed lib reference from soft to soft3d
1 lines of code changed in:
Temporary hack to work around DD_FALSE being undefined.
Fixed NT4 build. VC expect to see warnings until one of the VC
developers can manage something more elegant.
1 lines of code changed in:
Modifed to enable user to decide whether or not Trocchus'
DO_DINPUT_KEYBOARD should be enabled based
on compiler.
if COMP_VC & COMP_BC compilers being used, then default
is to enable Trochhus' DO_DINPUT_KEYBOARD routine.
4 lines of code changed in:
COMP_GCC specific modifications:
-- Disabled DO_DINPUT_KEYBOARD for GCC/Mingw/Cygwin32
3 lines of code changed in:
Mingw Specific Modifications:
-- Added options to decide which processor you want to use
Default is ix386
If you wish to use other Intel processors you will need to
modify CFLAGS.GENERAL
25 lines of code changed in:
fixed mixup of glxclean and glx2dclean rules in makefile
2 lines of code changed in:
Added reference to new csfont plugin.
2 lines of code changed in:
Now compiles with NextStep compiler.
1 lines of code changed in:
Updated dependency for dungeon.
1 lines of code changed in:
Received a patch from Samuel Humphreys which allows you to render
on texture dynamically. Currently this only works for the software
renderer. Support for the other renderers will have to be added.
There is also a new 'dtsimple' demo app to show of the new
feature.
3344 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