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: 2000-10-20 to 2000-10-10 (Commits 3595-3644 of 12744)
Made the following changes to the installation target:
-*- csengine.mak is no longer responsible for vfs.cfg and scf.cfg.
-*- Installation of vfs.cfg is now controlled by vfs.mak.
-*- Installation of scf.cfg is now controlled by csutil.mak (since SCF is
currently part of csutil).
27 lines of code changed in:
Set minimum canvas size for X to 32x32 instead of 320x200.
5 lines of code changed in:
Fix to my previous fix regarding Win32 gcc port
1 lines of code changed in:
Reverted back my previous change to vfs.mak; instead I have fixed
win32gcc.mak. DO.SHARED.PLUGIN should contain a reference to $(LIBS),
however win32gcc's DO.SHARED.PLUGIN did not.
11 lines of code changed in:
Added ZLib to the command-line for linking VFS plugin. Hmm, strange, how it
was able to link on other platforms before, especially on OS/2 which doesn't
permit unresolved symbols in dynamic libraries? Black magic.
2 lines of code changed in:
Fixed the dependency file name from "waveout.dep" to "wos.dep"
1 lines of code changed in:
Suddenly cswstest requires an extra CSGEOM.LIB dependency AFTER
CSSYS.LIB. Otherwise I got linking errors.
1 lines of code changed in:
Update.
48 lines of code changed in:
-*- Fixed a small bug in csRect::Exclude() that was there for a very long time.
Effectively it caused CSWS to do some extra work when resizing windows by
dragging their bottom margin.
-*- Fixed a bug in iG2D::DrawPixmap() when drawing alpha-transparent non-tiled
textures.
-*- Found and eliminated a clash between ex-cscmdRedraw and cscmdAreYouDefault.
Ooops. I believe this caused a lot of unneeded redraws before.
-*- Changed substantially the clipping subsystem in CSWS. Now it could be even
faster, besides it allows now semi-transparent windows to be repainted
correctly. Previously it did not care how many times the window redraws;
not it allows every dirty window to repaint exactly once, thus excluding
visual artifacts that existed before with semitransparent components.
The current clipping algorithm is much more sophisticated and tries
hardly to minimize the number of dirty regions; it tests for much more
complex cases (e.g. a opaque button in a transparent window inserted into
another component covering, say, a button inserted into desktop will
completely avoid redrawing the desktop button).
-*- Fixed all the X11 drivers so that the "csmcNone" mouse cursor is really
empty. Previously it was an uninitialized 1x1 pixmap, which sometimes
was visible like a black pixel.
-*- Had a flash about making clipped drawing in CSWS a lot faster. Now before
every Draw() method being called, the Redraw() method caches a "visible
region" which is effectively just a set of rectangles. Then every draw
operations just clips against this set of rectangles (if we're lucky,
we'll have just one rectangle in the visible region (e.g. for top windows),
thus doing just one clip-against-rectangle operation) rather than clipping
every graphic primitive against all childs, neightbours, parent's
neightbours and so on in the window hierarchy. MazeD should gain a lot
from this since for complex models it performs a lot of line and pixel
drawing.
As a side effect any drawing outside the ::Draw() method will not work
(it will be "clipped away" because the visible region cache will be empty
at that time). This is not a problem, and if turns to be so it could be
solved (but I would prefer not to).
504 lines of code changed in:
Re-added another entry of mine which Jorrit's Evil Twin deleted.
7 lines of code changed in:
Update.
18 lines of code changed in:
Fixed 3ds2lev texture mapping. I think it now works correctly.
2 lines of code changed in:
Fixed small typo.
1 lines of code changed in:
Updated out-of-date references to Win32 resources.
8 lines of code changed in:
Fixed out-of-date URL in map2cs tutorial.
1 lines of code changed in:
Updated out-of-date URL.
4 lines of code changed in:
-*- Fixed a number of mark-up problems and other errors in csws.txi.
Also eliminated 'overfull hbox' errors reported by TeX.
-*- Fixed a Texinfo error in beos.txi which was reported by 'makeinfo'.
-*- Fixed some out-of-date information and discussion in simple.txi.
193 lines of code changed in:
Obsolete.
0 lines of code changed in:
Further work on LevTool. Started work on portalization.
Added some new functions for edge/face information to csBox3.
222 lines of code changed in:
New csEventCord files.
65 lines of code changed in:
Big ol' CrystalClear update. Most of it is related to moving CC away from
impulses and towards using csEvents. csEventCords is in this update as well.
cctest input changes are in this commit as well.
107 lines of code changed in:
added a curve_templates member to the world. parser now saves the loaded templates there (until now it forgot about it and never freed memory for them)
8 lines of code changed in:
solved the problem with the 12 remaining instances of colldet at exit.
(when a thing deletes its attached objects it calls their destructors, wentity destructor itself removed _this_ object from parent and shifted others down. )
1 lines of code changed in:
closed more leaks
8 lines of code changed in:
closed 2 memory leaks (playing with the CS memory debugger the first time :)
7 lines of code changed in:
added initialization of BackTex member, now cswstest works again (window background look good)
4 lines of code changed in:
forgot to commit
2 lines of code changed in:
Fixed the 'clean --> cleandoc' chain rule in docs.mak which broke on
September 29, 2000.
6 lines of code changed in:
Maintenance changes to the BeOS port:
-*- Repaired so that it once again compiles and works following the many
core-level changes to the project over the last three months when this
port was idle. Now works with BeOS 5.
-*- Worked around a compiler optimization bug caused by 'schedule-insns2'
which resulted in QInt() returning completely bogus values when called
from csGraphics2D::DrawLine(). This bug crashed any program which
drew lines, such as g2dtest, but was particularly damaging to CSWS.
-*- Eliminated iBeLibSystemDriver (icsbe.h) interface. BeOS system driver
now implements iSystem::SystemExtension() instead. All communication
from plug-in modules is now done via SystemExtension() rather than
iBeLibSystemDriver.
-*- Works with new iEventPlug/iEventOutlet mechanism. Implements
iSystem::GetPotentiallyConflictingEvents() and QueryEventPriority().
The system driver now instantiates an iEventOutlet specifically for
keyboard and mouse events generated by the canvasses. These events
are handed to the system driver from canvasses via SystemExtension().
-*- Rewrote keyboard handling to work with Andrew's new "International
Keyboard Encoding" support.
-*- Added the new findlib.cpp and the newly mutated getopt.cpp from
libs/general/cssys to SRC.SYS_CSSYS in beos.mak.
Semi-major overhaul to BeOS port:
-*- Re-organized the threading model. Previously, Crystal Space would run
in the main thread until the first time iSystem::Loop() was called.
At that point, the BeOS application loop began running in the main
thread, and Crystal Space was shuffled over to a sub-thread. This was
a somewhat complex model. Under the new implementation, Crystal Space
always inhabits the main thread and the BeOS application loop runs in
the sub-thread.
-*- Furthermore, the BeOS application thread is now spawned and
BApplication::Run() invoked much earlier. Previously this task
occurred when iSystem::Loop() was first called. Now the application
thread is spawned and run as soon as the first window is created.
This allows the application and window to begin responding to events,
such as window-exposed events, even before Crystal Space has entered
its own run-loop. Canvasses now request a "BeginUI" extension via
iSystem::SystemExtension() to trigger the invocation of the BeOS
application's run-loop.
-*- Access to the Crystal Space event queue is now entirely thread-safe.
Previously, the event queue was accessed from both the Crystal Space
thread and the BeOS application thread. All messages which are
generated by the BeOS application thread are now placed into a
thread-safe message queue which is processed by the Crystal Space
thread in SysSystemDriver::NextFrame(). This applies to mouse and
keyboard events as well as other pseudo-events such as those generated
by the closing of a window or an application termination request.
-*- Eliminated the custom BApplication subclass (CrystApp). All
functionality which was previously handled by CrystApp is now handled
directly by SysSystemDriver. The system driver now also inherits from
BHandler and registers itself as the BApplication's preferred message
handler. This allows the system driver to receive and act upon the
standard BeOS messages without having to employ a custom BApplication
subclass.
-*- SysSystemDriver now only instantiates a BApplication if one has not
already been instantiated. Along with the obsolescence of CrystApp,
this allows BeOS-specific Crystal Space clients to easily employ
custom BApplication subclasses rather than being locked into CrystApp,
which was entirely private.
-*- Fixed the BeOS 2D canvas and Glide canvas so that they implement
BView::Draw(). This gives them the ability to redraw any portion of
the window when requested by BApplication, such as on a 'expose' event
when a portion of the window which was occluded becomes exposed. The
OpenGL canvas always had this capability.
-*- Canvasses now respect the 'area' argument passed to
iGraphics2D::Print().
-*- BeOS-specific csosdefs.h now manually works around problem with
redefinition of MAX() and MIN() in <sys/param.h> rather than relying
on source code to include headers in proper order. This was needed on
account of csbe.h now including BeOS headers since the BeOS system
driver inherits from BHandler and directly embeds a BMessageQueue.
-*- Removed the BeOS-specific "anatomy" discussion from the Crystal Space
Manual (anatomy/platform/beos.txi). This document was originally
written by David Huen to explain some of the operations of the BeOS
system driver and the BApplication subclass CrystApp. However,
following the current overhaul, the document was no longer accurate in
any way. In fact, its accuracy had slowly eroded over time, anyhow.
Moved the Bugs and To-Do lists from this document into the BeOS
installation document (build/platform/beos.txi).
BeOS Glide changes:
-*- Finally brought the BeOS Glide 2D driver (canvas) up to speed (as much
as possible). This driver has languished in a completely broken state
for a very long time, and has certainly been broken since my original
major overhaul of the BeOS port some time ago. The driver is untested
since I do not have the proper hardware, but it does compile.
-*- Fixed several problems with the BeOS Glide canvas makefile. Also
fixed the 'depend' target. Fixed the BeOS system makefile to
automatically configure Glide if present on the development system.
-*- Unified the SCF name of BeOS Glide canvas to match the format of names
of other Glide canvasses. Added 'glidebe2d' to scf.cfg.
-*- Updated documentation to reflect that the Glide driver now compiles
but is untested. Previously, documentation only stated that it was
entirely broken.
Common Glide changes:
-*- Eliminated compilation warnings in glide/glcache.cpp and
glide/glhalo.cpp about variables used uninitialized.
-*- Fixed Glide renderer makefiles so that it is possible to specify an
alternate Glide library. This feature is used by the BeOS port.
-*- Fixed Glide renderer makefiles so that they take the makefile variable
$(OUTDLL) into account when computing the DLL name.
-*- Fixed out of date Glide entries in scf.cfg. Added missing Glide3 2D
entry.
-*- Removed BeOS-specific garbage from the common Glide canvas
implementation.
211 lines of code changed in:
Eliminated compilation error (type mismatch between uint8 and UByte).
3 lines of code changed in:
-*- Reverted Jorrit's change to csVector::Get(). I (Eric) had already
removed the offending CS_ASSERT() call in the previous revision of
csvector.h. Jorrit actually removed the wrong one. It could not have
been the source of the problem he was experiencing since Get() is a
'const' method and does not alter the array at all. If Jorrit's code
was indeed using this 'const' method to access beyond the end of the
array, then that represents an actual bug in the calling code, and
needs to be fixed. The Get() method has never automatically extended
the array for out-of-bounds indexes; only the non-const version of
operator[] does so.
-*- Updated the class comments in csvector.h.
24 lines of code changed in:
Finished implementing the attribute System, including all
container classes specified in icsgame. (Not tested though)
2 lines of code changed in:
- Disabled a new CS_ASSERT that was put in csVector::Get(). The
problem was that some code in CS assumes that Get() also extends
the array. It is a bit too complicated to fix this now (I'm busy
fixing another problem and don't want to get sidetracked).
- Renamed csWorld::PrepareParticleSystems() to PrepareSprites() and
also changed the routine so that it updates all sprites and not
only particle systems.
- Fixed a bug as reported by Samuel Humphreys. Sprites that are
in nodes containing no geometry (i.e. no polygons) are not correctly
marked visible. The reason is that there is no bsp tree for those
nodes and the bsp tree is needed to hold the visibility stubs for
sprites. I solved this by creating a dummy bsp tree (empty) at
those nodes. Note that this will require a -recalc of levels
using STATBSP. This also holds for particle systems.
- Samuel also reported an additional bug. Sprites that are outside
the entire octree are never reported as visible. The reason is
similar, there is no bsp tree there to hold the visibility stubs.
The solution in this case is not so simple so for now I suggest
a workaround: just add dummy small polygons to the bounds of the
world where you want sprites to be visible. This will make the
octree larger.
34 lines of code changed in:
Update.
43 lines of code changed in:
-*- Renamed csPixmap::DrawTiledShifted into DrawTiled. There is no need to use
long clumsy names like DrawARoundBoxWithALittleRedDotInLeftUpperCorner ().
-*- I have extended iGraphics3D::DrawPixmap() method with one argument -
uint8 alpha, and also enhanced the function a lot in software renderer.
Now if the tx,ty,tw,th rectangle exceeds texture size, texture is tiled
across the screen rectangle. The alpha is from 0 to 255 and defines the
transparency of the pixmap, 0 being opaque and 255 full transparency.
-*- Semi-transparent windows in CSWS! (cswstest -> Theme button). This is a
bit buggy though because sometimes the window gets repainted twice, with
alpha-blended textures this results in visual artifacts :-( Will take
a look at this later.
Both windows and dialogs are allowed to be transparent now. Also simple
non-transparent textured backgrounds (which are faster of course) are
allowed for both windows and dialogs. Note that if you insert a dialog
into a window, and set both to semi-transparent, their textures will
overlap, and thus you will see the desktop (and the windows behind)
through *two* transparent textures. In general it is recommended that
if you insert a semi-transparent dialog into a window, you set that
window's alpha to 255 (e.g. fully transparent).
-*- Removed the NULL_SOFTWARE_2D_DRIVER macros from null3d renderer.
The NULL3D renderer is not a NULL renderer, it is a *NULL 3D* renderer,
that is, it does not have any 3D primitives implemented (e.g. polygons) -
everything other is like in the "normal" renderer, thus it uses same
default canvas like the normal renderer. NULL3D renderer is mainly an
alternative renderer for those CSWS applications that do not need any
3D drawing. Now it supports semi-transparent and tiled textures as well.
-*- Fixed other renderers in order to be able to compile. I believe tiling
will work with at least the OpenGL driver without any change; alpha
is not implemented yet but its easy to implement.
738 lines of code changed in:
Eliminated compilation warnings about uninitialized variables.
1 lines of code changed in:
Fixed several copmilation errors following Andrew's changes from yesterday.
2 lines of code changed in:
- Some small changes from Michael Robellard <miker@vetechno.com>:
- New SetValue in csKeyValuePair.
- Added 'const' to csMovable::Transform.
14 lines of code changed in:
xwindow is now destroyed after gl context is destroyed (other way around results in X error)
6 lines of code changed in:
Further work on leafy BSP generator in levtool.
2 lines of code changed in:
Removed some unneeded dependency stuff from unixconf.sh.
0 lines of code changed in:
Cleaned up formatting for the code from Brandon.
342 lines of code changed in:
Uncomment the MotionManager line in here to load it in
2 lines of code changed in:
There are some issues with frames where no interpolation occurs,
but Quaternion based skeletal animation is in and amazingly fast
even before attempting to optimize. Matrix mode animation is
supported, but no interpolation code exists yet. Tons of changes
to the csQuaternion class as it appears that nobody has used it yet
and one function added to csMatrix3 to support Quaternion
initialization. Also revamped the iMotion interfaces a bit, and
added a hacky SCF implementation for the csSkeleton classes until
someone with better knowledge can fix them so they are correct.
625 lines of code changed in:
Bug Fix:
Dropped "const" off "const char* filename = NULL;"
1 lines of code changed in:
This is the ex-"font manager" plug-in.
347 lines of code changed in:
Renamed "fontmanager" into "font server multiplexor". Moved the plugin
to another directory, re-registered with scfreg.
4 lines of code changed in:
A few fixes.
2 lines of code changed in:
Update.
67 lines of code changed in:
-*- Fixed makedep so that it will work unde Unix (well, at least under Linux :-)
Added autodetection of system include directories to unixconf.sh.
You will have to re-configure in order "make depend" to work.
Added a new AWK script bin/includes.awk which parses the output of
"gcc -H" and outputs a line like this:
MAKEDEP.INCLUDES=-I/usr/include;/usr/local/include
Maybe I'll enable later this script for mingw32 port as well, since
mingw32 needs these includes too; currently you have to specify them
as "C_INCLUDE_PATH" environment variable.
-*- Changed Unix makefile so that makedep is used now on Unix for dependency
generation. This is about five times faster (a random evaluation though :-)
-*- Re-implemented former `themes' from scratch. It was a nice try, but did not
fit well with CSWS concept; besides they involved too much overhead.
They are called now "skins". A "skin" is a collection of "skin slices",
every "slice" is responsible for a certain class of components, e.g.
buttons, titlebars and so on. When a component is created or when the
skin is changed the component queries the skin for the slice it wants
and remembers it; then it uses it for miscelaneous operations such as
drawing and so on.
There are three skinnable components right now: buttons, windows and
dialogs. There is just one (default) scheme right now thus you won't
notice any difference :-) The code for exterior look of these components
is completely separated (and can be replaced by other code, e.g. Motif,
Windoze, OS/2 look (um... the default look is quite close to OS/2
already :-) etc).
-*- Changed CSWS configuration file format. Now it use csIniFile. This is a
preparation so that themes could be easier configured through config
file in future.
-*- Updated CSWS docs so that it can be used to gather basic information about
CSWS. All beginner CSWS users are highly recommended that; I have tried
to respond to the most frequently asked questions on the mailing list
there.
-*- Renamed csApp::InitialSetup() into csApp::Initialize(const char *iConfigName)
for uniformity (all other classes uses Initialize()). I have fixed the
applications in CVS, for applications outside CVS this have to be done
yet. The iConfigName passed is the name of CSWS configuration file
(usually "/csws/csws.cfg"). This allows to create CSWS applications
without having a system configuration file (previously CSWS config
name was read from the [CSWS] section of system config file).
-*- Added the notion of "color schemes". There is a function called
csSetColorScheme() which in fact applies a (tricky) function onto the
entire application palette (in HLS space). There are just too many colors
in all palettes to enumerate them separately; I would prefer some automatic
way to process all colors in the palette; I'm not sure the method that
is currently implemented is flexible enough (I'm going to finish the
"Theme" window in cswstest in order to see schemes in action dynamically).
-*- Fixed a small bug in CSWS that caused transparent windows to leave visible
traces on neightbour's child components. The solution was simple and
elegant - just a single line change! :-) Now you can even drag a window
*below* the transparent window (e.g. try to drag the "CSS_TOPSELECT"
window under the "Theme test" window in cswstest!). This looks cool,
although it will look much better when I will implement support for
alpha backgrounds! :-)
-*- Removed the #ifdef MSC_VER from csapp.cpp. Grrrr.... such things make
me angry.
2126 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