You can subscribe to this list here.
2001 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
(52) |
Nov
(77) |
Dec
(5) |
---|---|---|---|---|---|---|---|---|---|---|---|---|
2002 |
Jan
(21) |
Feb
(12) |
Mar
(49) |
Apr
(20) |
May
(7) |
Jun
(41) |
Jul
(39) |
Aug
(49) |
Sep
(95) |
Oct
(148) |
Nov
(131) |
Dec
(73) |
2003 |
Jan
(195) |
Feb
(113) |
Mar
(84) |
Apr
(52) |
May
(100) |
Jun
(89) |
Jul
(124) |
Aug
(94) |
Sep
(107) |
Oct
(195) |
Nov
(75) |
Dec
(64) |
2004 |
Jan
(126) |
Feb
(99) |
Mar
(118) |
Apr
(96) |
May
(176) |
Jun
(119) |
Jul
(139) |
Aug
(158) |
Sep
(198) |
Oct
(74) |
Nov
(195) |
Dec
(93) |
2005 |
Jan
(205) |
Feb
(201) |
Mar
(197) |
Apr
(216) |
May
(227) |
Jun
(239) |
Jul
(173) |
Aug
(179) |
Sep
(227) |
Oct
(271) |
Nov
(225) |
Dec
(167) |
2006 |
Jan
(160) |
Feb
(188) |
Mar
(242) |
Apr
(140) |
May
(229) |
Jun
(154) |
Jul
(207) |
Aug
(385) |
Sep
(445) |
Oct
(381) |
Nov
(249) |
Dec
(178) |
2007 |
Jan
(187) |
Feb
(188) |
Mar
(346) |
Apr
(177) |
May
(254) |
Jun
(234) |
Jul
(112) |
Aug
(198) |
Sep
(181) |
Oct
(183) |
Nov
(132) |
Dec
(57) |
2008 |
Jan
(70) |
Feb
(139) |
Mar
(129) |
Apr
(127) |
May
(143) |
Jun
(112) |
Jul
(71) |
Aug
(162) |
Sep
(89) |
Oct
(114) |
Nov
(242) |
Dec
(130) |
2009 |
Jan
(175) |
Feb
(154) |
Mar
(299) |
Apr
(99) |
May
(117) |
Jun
(240) |
Jul
(173) |
Aug
(156) |
Sep
(131) |
Oct
(71) |
Nov
(102) |
Dec
(42) |
2010 |
Jan
(112) |
Feb
(102) |
Mar
(128) |
Apr
(91) |
May
(116) |
Jun
(114) |
Jul
(105) |
Aug
(49) |
Sep
(40) |
Oct
(35) |
Nov
(43) |
Dec
(46) |
2011 |
Jan
(58) |
Feb
(97) |
Mar
(40) |
Apr
(55) |
May
(107) |
Jun
(33) |
Jul
(20) |
Aug
(27) |
Sep
(73) |
Oct
(77) |
Nov
(84) |
Dec
(48) |
2012 |
Jan
(51) |
Feb
(46) |
Mar
(52) |
Apr
(37) |
May
(19) |
Jun
(24) |
Jul
(46) |
Aug
(62) |
Sep
(42) |
Oct
(33) |
Nov
(56) |
Dec
(26) |
2013 |
Jan
(29) |
Feb
(45) |
Mar
(36) |
Apr
(18) |
May
(51) |
Jun
(13) |
Jul
(6) |
Aug
(17) |
Sep
(11) |
Oct
(37) |
Nov
(10) |
Dec
(6) |
2014 |
Jan
(24) |
Feb
(11) |
Mar
(41) |
Apr
(14) |
May
(34) |
Jun
(33) |
Jul
(27) |
Aug
(30) |
Sep
(18) |
Oct
(18) |
Nov
(20) |
Dec
(4) |
2015 |
Jan
(6) |
Feb
(8) |
Mar
(12) |
Apr
(3) |
May
(4) |
Jun
(6) |
Jul
(2) |
Aug
(8) |
Sep
|
Oct
(20) |
Nov
(11) |
Dec
(10) |
2016 |
Jan
(5) |
Feb
(2) |
Mar
(4) |
Apr
(3) |
May
(8) |
Jun
(6) |
Jul
(5) |
Aug
(1) |
Sep
(10) |
Oct
(6) |
Nov
(5) |
Dec
(1) |
2017 |
Jan
(22) |
Feb
(16) |
Mar
(10) |
Apr
|
May
(14) |
Jun
(1) |
Jul
(1) |
Aug
|
Sep
|
Oct
|
Nov
(8) |
Dec
|
2018 |
Jan
|
Feb
|
Mar
|
Apr
|
May
(5) |
Jun
|
Jul
|
Aug
(4) |
Sep
(1) |
Oct
|
Nov
|
Dec
(2) |
2019 |
Jan
(1) |
Feb
(1) |
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
(3) |
Dec
(3) |
2020 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
(9) |
2021 |
Jan
|
Feb
(7) |
Mar
(7) |
Apr
(2) |
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
From: Dirk R. <re...@ig...> - 2002-03-19 09:50:14
|
I can't answer all of your questions, Gerrit is the thread master, he'll cover the rest. On Mar 18, 6:48pm, Joerg Treschau (S. Schneider) wrote: > Subject: [Opensg-users] Threads (was Re: Deleting old Nodes) > > How are aspects handled and how are they created? > i tried to run a small test program, using 4 aspects (1, 2, 3, 4). for 2, 3 > and 4 i got the following error: > FATAL: OSGPTB : invalid aspect id > why? how can this be avoided? The maximum number of aspects has to be set before osgInit is called, as we need them to allocate the structures and changing it after init would necessitate reallocating all existing objects, which we can't really do, as we would have to change all pointers to objects. Right now it's compiled in to be 2, where aspect ids start at 0. If you need more you can change it in Base/OSGConfig.h. Support for changing it at runtime is experimental, Gerrit can tell you more about it. > when i understand aspects right, then it should be the best to only use one > aspect, when changing separated scenegraph objects from different threads. > is this correct? If you can ensure that they don't interfere yes. Changes to scenegraph nodes can influnce other nodes, primarily by invalidating lazily evaluated data like bounding volumes. If you only invalidate stuff I don't see any problems, if you want to validate stuff (e.g. update the volume) in parallel you can run into problems. If you're creating nodes in parallel adding them to the same parent will be problematic. Write access to structures is not locked in OpenSG, as locking and synchronisation reduce threading benefits. I guess you want to create nodes in parallel and combine them to a single graph. The best way to do that is creating independent subgraphs in the threads and adding them to the root in the master thread. >-- End of excerpt from Joerg Treschau (S. Schneider) Hope it helps Dirk -- -- -- Dirk Reiners OpenSG Forum di...@op... -- The OpenSG Open Source Scenegraph: http://www.opensg.org -- Announcements: http://lists.sf.net/lists/listinfo/opensg-announce |
From: Joerg T. (S. Schneider) <tre...@ig...> - 2002-03-18 17:48:41
|
>> I had expected something like with qt threads: a thread class to inherit >> with a virtual run method that has to be redefined. > Threads are handled differently everywhere. A Thread class with a virtual run method is at least what Qt and Java use as object-oriented representation of Threads, so i had expected something similar with OpenSG. IMHO i like the way Qt and Java abstract the use of threads much more than the way OpenSG does. it 'feels' more object oriented. I still have problems using OpenSG with multiple threads. the code looks somewhat like that: thread = Thread::get(name); if(thread == NULL) { cerr << "error: could not create thread!" << endl; return; } thread->run(function, aspect, (void*) param); This little piece of code leads many questions: The first is: why do i have to create a name for every thread? i dynamically create new threads relatively often, so this has many implications that arise: can i call Thread::get(name) more than once for the same name, or do i have to create a new name for every new thread? is it safe when i know the old thread with the same name hase ended? is it safe when the old thread has not ended yet? what happens to old threads. are they removed from the store somewhen or do they stay there forever? staying forever would be a major issue when i have to dynamically creating a new name for every new thread. How are aspects handled and how are they created? i tried to run a small test program, using 4 aspects (1, 2, 3, 4). for 2, 3 and 4 i got the following error: FATAL: OSGPTB : invalid aspect id why? how can this be avoided? when i understand aspects right, then it should be the best to only use one aspect, when changing separated scenegraph objects from different threads. is this correct? Thanx Joerg |
From: Joerg T. (S. Schneider) <tre...@ig...> - 2002-03-18 17:43:51
|
> If you're only changing data and not structure I'd recommend not recreating > Nodes all the time, just changing existing nodes is definitely more > efficient. No idea what exactly you're doing, though. i am recreating geometry in various nodes, possibly creating more or less nodes. changing the geometry of existing nodes is no real solution for me. > In 1.0 you'll have to use node->unlinkSubTree(); to remove all the children > and subRefCP(node); to remove the node. i have troubles using this method for deletion of old nodes. my application crashes after i delete an indetermistic number of nodes this way. i have changed tutorial 05 to test this deletion method and it crasches too after deleting approx. 3 to 7 nodes. the changed source is attached. can anyone test it please. with space the node is replaced with a new one, containing new geometry that looks the same. >> i tried some cvs snapshots, but was never able to use (compile) one >> successfuly. are there any working cvs versions?? > Hmm, the dailybuild compiles them every night, pretty successfully. Which > platform, and what exactly were the problems? the last snapshot i tried to use was OpenSG-020302-src. The problem: /opt/OpenSG-020302-src/Experimental/Text/OSGVectorFontGlyph.cpp: In method `bool osg::VectorFontGlyph::createTriangles ()': /opt/OpenSG-020302-src/Experimental/Text/OSGVectorFontGlyph.cpp:143: `gluBeginPolygon' undeclared (first use this function) btw. my system is a linux box with gcc 2.96-0.76mdk (build date Thu 21 Feb 2002) and glibc 2.2.4. i know, that gcc 2.96 may be problematic, but i never experienced any problems so far. Thanx Joerg |
From: Dirk R. <re...@ig...> - 2002-03-17 14:14:55
|
On Mar 8, 2:54am, Erik Ferner wrote: > Subject: [Opensg-users] Problems accessing some things in OSGNavigator > Hi, > > I'm trying to create some Hotkeys in my Application to apply preset > views. Like CTRL-1 is a front view, CTRL-2 a side view from the left , > CTRL-3 from the right and so on... > Nothing really tricky, but I've got problems finding the methods to do > this. Basically I just need to set a position from where I'm looking and > a position to where I'm looking. I can get the OSGNavigator from my > SimpleSceneManager, but I can't access any methods to do this. Here's > what I've tried: > > Since the FlyNavigator is private, I can't access its methods 'setFrom' > and 'setAt', which would be exactly what I needed. > I can get the matrix of the FlyNavigator but changing it doesn't have > any effect. That matrix is not what's used inside, so changing it doesn't help. In general it's a little trickier, as the Navigator hides a FlyNavigator and a TrackballNavigator, only one of which is going to be active at any given time. So changing only the FlyNavigator doesn't really help much. > Will there be some corresponding methods in OSGNavigator in the near > future, or is there another way ? > >-- End of excerpt from Erik Ferner I can't think of another way, we'll add them. Yours Dirk -- -- -- Dirk Reiners OpenSG Forum di...@op... -- The OpenSG Open Source Scenegraph: http://www.opensg.org -- Announcements: http://lists.sf.net/lists/listinfo/opensg-announce |
From: Dirk R. <re...@ig...> - 2002-03-15 12:09:15
|
On Mar 13, 6:35pm, Kai-Peter B=E4ckman wrote: > Subject: [Opensg-users] Labels > > Hello, > > Is there a way to make labels (ie. billboard text) using OpenSG? And w= hat is > the correct way at the moment? I am using the 0310 dailybuild > >-- End of excerpt from Kai-Peter B=E4ckman There's no easy way right now, as the Text lib is not quite integrated ye= t. There's no text node and the library doewn't compile on Windows yet. If you need it and you're on Unix you can use the Text lib, see Experimental/Text/testTextRender.cpp for an example on how to use it. It = can create the text in an image (to be used as a single texture), as 3D geome= try or as a texture/geometry pair for fast changing textured text. You can put the result into a Billboard for a label. If you have lots of labels it might be a little slow, an alternative woul= d be to write a Label note for that, but that's quite a bit more work. Hope it helps Dirk -- = -- -- Dirk Reiners OpenSG Forum dirk@opensg.or= g = -- The OpenSG Open Source Scenegraph: http://www.opensg.or= g -- Announcements: http://lists.sf.net/lists/listinfo/opensg-announc= e |
From: Gerrit V. <vo...@ca...> - 2002-03-15 10:29:29
|
Hi, Christian Henning wrote: > ----- Original Message ----- > From: "Gerrit Voss" <vo...@ca...> > To: <ope...@li...> > Sent: Friday, March 15, 2002 9:04 AM > Subject: Re: [Opensg-users] Visual c++ > > > >>Hi, >> >>Christian Henning wrote: >> >> >>>Hi all, >>> >>> >>> >>>after I downloaded the cvs tree of OpenSG. >>> >> >>cvs checkout or download from the dailybuild page ? >>If cvs windows or unix ? >> > > Yesterday I made checkout via cvs on linux. Oh, now I can imagine the > problem!! Some minutes ago I downloaded the source from the dailybuild page > using IE on Windows 2000 and everything is working. OK, it was my fault. > Sorry > > The linefeed is different on UNIX-like environments and so every line seems > very long and unreadable. > Yep these make the small difference, it's one of those tricks you learns over time ;-). > >>>I was going to open the >>>workspace file in the directory VsBuild/, but nothing happend. >>> >> >>is that no project appeared ? >> >> >>>NoMessage or something. Then I build my own version with cygwin using >>>"make dsp" and again nothing happend with Visual, after opening the .dsw >>>file. The .dsp's file in the subfolder are created on my computer. >>> >>> >> >>Can you send me the .dsw and the Base.dsp file, then I can have a closer >>look at it. >> > > Do you still want the .dsw and Base.dsp file? No, only if something goes wrong, but if you use the ones from the dailybuild page, I can grep them myself in case of trouble. Gerrit |
From: Christian H. <he...@ro...> - 2002-03-15 10:02:39
|
----- Original Message ----- From: "Gerrit Voss" <vo...@ca...> To: <ope...@li...> Sent: Friday, March 15, 2002 9:04 AM Subject: Re: [Opensg-users] Visual c++ > > Hi, > > Christian Henning wrote: > > > Hi all, > > > > > > > > after I downloaded the cvs tree of OpenSG. > > > cvs checkout or download from the dailybuild page ? > If cvs windows or unix ? Yesterday I made checkout via cvs on linux. Oh, now I can imagine the problem!! Some minutes ago I downloaded the source from the dailybuild page using IE on Windows 2000 and everything is working. OK, it was my fault. Sorry The linefeed is different on UNIX-like environments and so every line seems very long and unreadable. > > > I was going to open the > > workspace file in the directory VsBuild/, but nothing happend. > > > is that no project appeared ? > > > NoMessage or something. Then I build my own version with cygwin using > > "make dsp" and again nothing happend with Visual, after opening the .dsw > > file. The .dsp's file in the subfolder are created on my computer. > > > > > Can you send me the .dsw and the Base.dsp file, then I can have a closer > look at it. Do you still want the .dsw and Base.dsp file? > > Gerrit > Christian |
From: Gerrit V. <vo...@ca...> - 2002-03-15 08:06:55
|
Hi, Christian Henning wrote: > Hi all, > > > > after I downloaded the cvs tree of OpenSG. cvs checkout or download from the dailybuild page ? If cvs windows or unix ? > I was going to open the > workspace file in the directory VsBuild/, but nothing happend. is that no project appeared ? > NoMessage or something. Then I build my own version with cygwin using > "make dsp" and again nothing happend with Visual, after opening the .dsw > file. The .dsp's file in the subfolder are created on my computer. > Can you send me the .dsw and the Base.dsp file, then I can have a closer look at it. Gerrit |
From: Christian H. <he...@ro...> - 2002-03-14 21:14:40
|
Hi all, after I downloaded the cvs tree of OpenSG. I was going to open the = workspace file in the directory VsBuild/, but nothing happend. No = Message or something. Then I build my own version with cygwin using = "make dsp" and again nothing happend with Visual, after opening the .dsw = file. The .dsp's file in the subfolder are created on my computer. What is wrong? regards, Christian |
From: Kai-Peter <Kai...@mi...> - 2002-03-13 16:35:40
|
Hello, Is there a way to make labels (ie. billboard text) using OpenSG? And what is the correct way at the moment? I am using the 0310 dailybuild Kai-Peter Bäckman |
From: Gerrit V. <vo...@ca...> - 2002-03-13 01:19:11
|
Hi, yes it seems that the lock semantic is different on different platforms, I'll have a closer look at it. Gerrit Erik Ferner wrote: > Hi everybody, > > I'm trying to delete a single leaf node from the scenegraph by calling > > osg::NodePtr _Parent = leafNode->getParent(); > _Parent->subChild( leafNode ); > Sometimes it works fine and sometimes the application locks up and I > have to interrupt it by pressing CTRL-C. > The weird thing is that this only happens sporadically. > Here's the backtrace from the debugger after pressing CTRL-C: > > Program received signal SIGINT, Interrupt. > [Switching to Thread 1024 (LWP 9173)] > 0x41c46906 in sigsuspend () from /lib/libc.so.6 > (gdb) bt > #0 0x41c46906 in sigsuspend () from /lib/libc.so.6 > #1 0x41d477f0 in __pthread_wait_for_restart_signal () from > /lib/libpthread.so.0 > #2 0x41d496b5 in __pthread_alt_lock () from /lib/libpthread.so.0 > #3 0x41d45e72 in pthread_mutex_lock () from /lib/libpthread.so.0 > #4 0x40d1a146 in osg::PThreadLockBase::aquire (this=0x81e46d0) at > /home/homer/src/OpenSG-CVS/OpenSG/Base/OSGLock.cpp:119 > #5 0x40d1a5e4 in osg::LockPool::aquire (this=0x81e4440, keyP=0x82ec860) > at /home/homer/src/OpenSG-CVS/OpenSG/Base/OSGLock.cpp:502 > #6 0x404fbdad in osg::FieldContainerPtrBase::subRef (this=0x82ef8a0) > at > /home/homer/src/OpenSG-CVS/OpenSG/FieldContainer/OSGFieldContainerPtr.cpp:448 > > #7 0x404f8bcc in osg::subRefCP (objectP=@0x82ef8a0) at > /home/homer/src/OpenSG-CVS/OpenSG/FieldContainer/OSGFieldContainer.cpp:199 > #8 0x40501776 in osg::Node::~Node (this=0x82ef830, __in_chrg=2) at > /home/homer/src/OpenSG-CVS/OpenSG/FieldContainer/OSGNode.cpp:832 > #9 0x404fbea6 in osg::FieldContainerPtrBase::subRefUnlocked > (this=0xbfffe414) > at > /home/homer/src/OpenSG-CVS/OpenSG/FieldContainer/OSGFieldContainerPtr.cpp:482 > > #10 0x404fbdc9 in osg::FieldContainerPtrBase::subRef (this=0xbfffe414) > at > /home/homer/src/OpenSG-CVS/OpenSG/FieldContainer/OSGFieldContainerPtr.cpp:452 > > #11 0x404f8bcc in osg::subRefCP (objectP=@0xbfffe414) at > /home/homer/src/OpenSG-CVS/OpenSG/FieldContainer/OSGFieldContainer.cpp:199 > #12 0x405000d8 in osg::Node::subChild (this=0x8299700, childP=@0xbfffe414) > at /home/homer/src/OpenSG-CVS/OpenSG/FieldContainer/OSGNode.cpp:400 > #13 0x080a91a5 in InventorLoader::checkForRedundancy (this=0x8233f00, > OSGGroup= > {<AttachmentContainerPtr> = {<FieldContainerPtr> = > {<FieldContainerPtrBase> = {static InvalidParentEPos = 65535, > _containerSize = 58388, _uiParentEPos = 49151, _storeP = 0x82ecc40 > "\001", static _pRefCountLock = 0x81e4440}, <No data fields>}, <No data > fields>}, <No > data fields>}) at InventorLoader.cpp:491 > > Any ideas ? (I'm using the most recent CVS checkout) > > Thanx alot > Erik > > > _______________________________________________ > Opensg-users mailing list > Ope...@li... > https://lists.sourceforge.net/lists/listinfo/opensg-users |
From: Erik F. <fe...@st...> - 2002-03-12 18:43:05
|
Hi everybody, I'm trying to delete a single leaf node from the scenegraph by calling osg::NodePtr _Parent = leafNode->getParent(); _Parent->subChild( leafNode ); Sometimes it works fine and sometimes the application locks up and I have to interrupt it by pressing CTRL-C. The weird thing is that this only happens sporadically. Here's the backtrace from the debugger after pressing CTRL-C: Program received signal SIGINT, Interrupt. [Switching to Thread 1024 (LWP 9173)] 0x41c46906 in sigsuspend () from /lib/libc.so.6 (gdb) bt #0 0x41c46906 in sigsuspend () from /lib/libc.so.6 #1 0x41d477f0 in __pthread_wait_for_restart_signal () from /lib/libpthread.so.0 #2 0x41d496b5 in __pthread_alt_lock () from /lib/libpthread.so.0 #3 0x41d45e72 in pthread_mutex_lock () from /lib/libpthread.so.0 #4 0x40d1a146 in osg::PThreadLockBase::aquire (this=0x81e46d0) at /home/homer/src/OpenSG-CVS/OpenSG/Base/OSGLock.cpp:119 #5 0x40d1a5e4 in osg::LockPool::aquire (this=0x81e4440, keyP=0x82ec860) at /home/homer/src/OpenSG-CVS/OpenSG/Base/OSGLock.cpp:502 #6 0x404fbdad in osg::FieldContainerPtrBase::subRef (this=0x82ef8a0) at /home/homer/src/OpenSG-CVS/OpenSG/FieldContainer/OSGFieldContainerPtr.cpp:448 #7 0x404f8bcc in osg::subRefCP (objectP=@0x82ef8a0) at /home/homer/src/OpenSG-CVS/OpenSG/FieldContainer/OSGFieldContainer.cpp:199 #8 0x40501776 in osg::Node::~Node (this=0x82ef830, __in_chrg=2) at /home/homer/src/OpenSG-CVS/OpenSG/FieldContainer/OSGNode.cpp:832 #9 0x404fbea6 in osg::FieldContainerPtrBase::subRefUnlocked (this=0xbfffe414) at /home/homer/src/OpenSG-CVS/OpenSG/FieldContainer/OSGFieldContainerPtr.cpp:482 #10 0x404fbdc9 in osg::FieldContainerPtrBase::subRef (this=0xbfffe414) at /home/homer/src/OpenSG-CVS/OpenSG/FieldContainer/OSGFieldContainerPtr.cpp:452 #11 0x404f8bcc in osg::subRefCP (objectP=@0xbfffe414) at /home/homer/src/OpenSG-CVS/OpenSG/FieldContainer/OSGFieldContainer.cpp:199 #12 0x405000d8 in osg::Node::subChild (this=0x8299700, childP=@0xbfffe414) at /home/homer/src/OpenSG-CVS/OpenSG/FieldContainer/OSGNode.cpp:400 #13 0x080a91a5 in InventorLoader::checkForRedundancy (this=0x8233f00, OSGGroup= {<AttachmentContainerPtr> = {<FieldContainerPtr> = {<FieldContainerPtrBase> = {static InvalidParentEPos = 65535, _containerSize = 58388, _uiParentEPos = 49151, _storeP = 0x82ecc40 "\001", static _pRefCountLock = 0x81e4440}, <No data fields>}, <No data fields>}, <No data fields>}) at InventorLoader.cpp:491 Any ideas ? (I'm using the most recent CVS checkout) Thanx alot Erik |
From: Dirk R. <re...@ig...> - 2002-03-12 12:38:31
|
On Mar 12, 12:31pm, J=F6rg Treschau wrote: > Subject: [Opensg-users] Deleting old Nodes > Hi! > Seems, like i am not the only one having problems with old nodes > consuming memory. > > I am writing a scientific visualization app and must re-create geometry= > very often, so it ist VERY important for me that old nodes get deleted.= If you're only changing data and not structure I'd recommend not recreati= ng Nodes all the time, just changing existing nodes is definitely more efficient. No idea what exactly you're doing, though. > i use > groupNode->replaceChildBy(oldNode, newNode) > to replace the nodes, but no memory is ever freed, instead the memory > used by the process grows and grows. > > Is there some way to force the deletion of unused nodes? > >-- End of excerpt from J=F6rg Treschau In current CVS the reference counting has been redone, so that this shoul= d work now, AFAIK. Gerrit? In 1.0 you'll have to use node->unlinkSubTree(); to remove all the childr= en and subRefCP(node); to remove the node. On Mar 12, 12:37pm, J=F6rg Treschau wrote: > Subject: [Opensg-users] OpenSG Threads > Hi! > > Can anyone explain to me, how the OpenSG Threads classes should be used= ? > I have no clue so far. > I had expected something like with qt threads: a thread class to inheri= t > with a virtual run method that has to be redefined. > > The Starter-Guide says: oh yes, of course you can use multiple threads,= > but doesn't tell how and there is no example that shows how to use them= =2E > The documentation of the Thread classes is pretty non-existent, so it > didn't help me either. > > My first try, using qt threads resulted in a crash, cause the aspect ar= e > not set up by them... > >-- End of excerpt from J=F6rg Treschau Threads are handled differently everywhere. For a very simple example see= Experimental/tutorials/08parloading.cpp. Quick version: create a thread object and call its run() method with the function to execute in a separate thread. On Mar 12, 12:47pm, J=F6rg Treschau wrote: > Subject: [Opensg-users] Special rendering control > Hi! > > For a visualization method i program i need special control about the > rendering order and blending parameters. > I have to render several co-planar triangles with special blending settings. > How can this be achieved best with OpenSG? Blending is defined via a BlendChunk in the material. You can add chunks = to ChunkMaterial and descendents (Simple*Material). Rendering order can not be defined externally yet. We only do transparent= objects after opaque ones right now. Sorry, it's on the list, but not her= e yet. :( > I tried to create an own subclass of NodePtr for this purpose but faile= d > to do so. > Compiling the fcdEdit program from the 1.0-src release was not very > easy, because i had to set some paths by hand. Wy is it not included in= > the build-system? Because it was initially written for another system, and the OpenSG build= system is targetted to create the OpenSG library and examples, not standa= lone apps. That's why the examples and tutotials have their own makefiles, too= =2E I fixed the makefile in the current CVS build to work better, but you sti= ll need to set QTDIR (which most QT users have set anyway). > When I used fcdEdit to create the source-files for my own nodes it > always crashed when creating the source code for the classes. Running > the program from the command line with parameters to create the source > code from the given .fcd file it crashed too. Never happened to us before (at least not for versions that are checked i= n ;). Which platform, what is the .fcd you were using? > any suggestions? > i tried some cvs snapshots, but was never able to use (compile) one > successfuly. are there any working cvs versions?? Hmm, the dailybuild compiles them every night, pretty successfully. Which= platform, and what exactly were the problems? >-- End of excerpt from J=F6rg Treschau Dirk -- = -- -- Dirk Reiners OpenSG Forum dirk@opensg.or= g = -- The OpenSG Open Source Scenegraph: http://www.opensg.or= g -- Announcements: http://lists.sf.net/lists/listinfo/opensg-announc= e |
From: Gerrit V. <vo...@ca...> - 2002-03-12 12:14:31
|
Hi, J=F6rg Treschau wrote: > Hi! >=20 > Can anyone explain to me, how the OpenSG Threads classes should be used= ?=20 > I have no clue so far. > I had expected something like with qt threads: a thread class to inheri= t=20 > with a virtual run method that has to be redefined. That should be possible (IIRC). >=20 > The Starter-Guide says: oh yes, of course you can use multiple threads,= =20 > but doesn't tell how and there is no example that shows how to use them= .=20 > The documentation of the Thread classes is pretty non-existent, so it=20 > didn't help me either. >=20 > My first try, using qt threads resulted in a crash, cause the aspect ar= e=20 > not set up by them... This feature was added to the cvs version recently (called ExternalThread). >=20 I'll try to find a small example showing how to use it, unfortunately I am quite buisy with other stuff right now so it might take a while ;-(. Gerrit |
From: Gerrit V. <vo...@ca...> - 2002-03-12 12:12:51
|
Hi, if you are using 1.0 : yes it is broken there ;-( I fixed the refcount stuff recently for the current CVS version, but it still needs some testing. But most of the stuff gets deleted if the ref count drops to zero. Gerrit J=F6rg Treschau wrote: > Hi! > Seems, like i am not the only one having problems with old nodes=20 > consuming memory. >=20 > I am writing a scientific visualization app and must re-create geometry= =20 > very often, so it ist VERY important for me that old nodes get deleted. >=20 > i use > groupNode->replaceChildBy(oldNode, newNode) > to replace the nodes, but no memory is ever freed, instead the memory=20 > used by the process grows and grows. >=20 > Is there some way to force the deletion of unused nodes? >=20 > Joerg Treschau >=20 >=20 > _______________________________________________ > Opensg-users mailing list > Ope...@li... > https://lists.sourceforge.net/lists/listinfo/opensg-users |
From: T. <tre...@ig...> - 2002-03-12 11:46:32
|
Hi! For a visualization method i program i need special control about the rendering order and blending parameters. I have to render several co-planar triangles with special blending settings. How can this be achieved best with OpenSG? I tried to create an own subclass of NodePtr for this purpose but failed to do so. Compiling the fcdEdit program from the 1.0-src release was not very easy, because i had to set some paths by hand. Wy is it not included in the build-system? When I used fcdEdit to create the source-files for my own nodes it always crashed when creating the source code for the classes. Running the program from the command line with parameters to create the source code from the given .fcd file it crashed too. any suggestions? i tried some cvs snapshots, but was never able to use (compile) one successfuly. are there any working cvs versions?? Thanx Joerg |
From: T. <tre...@ig...> - 2002-03-12 11:36:42
|
Hi! Can anyone explain to me, how the OpenSG Threads classes should be used? I have no clue so far. I had expected something like with qt threads: a thread class to inherit with a virtual run method that has to be redefined. The Starter-Guide says: oh yes, of course you can use multiple threads, but doesn't tell how and there is no example that shows how to use them. The documentation of the Thread classes is pretty non-existent, so it didn't help me either. My first try, using qt threads resulted in a crash, cause the aspect are not set up by them... Thanx Joerg |
From: T. <tre...@ig...> - 2002-03-12 11:30:15
|
Hi! Seems, like i am not the only one having problems with old nodes consuming memory. I am writing a scientific visualization app and must re-create geometry very often, so it ist VERY important for me that old nodes get deleted. i use groupNode->replaceChildBy(oldNode, newNode) to replace the nodes, but no memory is ever freed, instead the memory used by the process grows and grows. Is there some way to force the deletion of unused nodes? Joerg Treschau |
From: Christian H. <he...@ro...> - 2002-03-11 18:42:57
|
sorry for the late reply, here are my compiler and linker options, taken from the project settings dialog: -- compiler settings -- /nologo /MDd /W3 /Gm /GR /ZI /Od /I "C:\Programme\OpenSG-1.0\include" /D "WIN32" /D "_DEBUG" /D "_WINDOWS" /D "_AFXDLL" /D "OSG_WITH_WIN32" /D WINVER=0x0400 /D _WIN32_WINDOWS=0x0410 /D "OSG_BUILD_DLL" /D "OSG_NEW_DLLS" /D "_OSG_HAVE_CONFIGURED_H_" /Fo"Debug/" /Fd"Debug/" /FD /GZ /c -- linker settings -- OSGBase.lib OSGSystem.lib OSGWindowGLUT.lib OSGWindowQT.lib OSGWindowWIN32.lib openGL32.lib winmm.lib /nologo /subsystem:windows /incremental:yes /pdb:"Debug/mfctest.pdb" /debug /machine:I386 /out:"Debug/mfctest.exe" /pdbtype:sept /libpath:"C:\Programme\OpenSG-1.0\lib" Please notice these are the settings for the debug version. Thanks, Christian ----- Original Message ----- From: "Gerrit Voss" <vo...@ca...> To: <ope...@li...> Sent: Friday, March 08, 2002 2:35 AM Subject: Re: [Opensg-users] strange linker error > > Hi, > > could you send me the complete build log (compile + link option) > so I can have a closer look at it. > > Thanks > Gerrit > > > Christian Henning wrote: > > > Thanks for answer. > > > > I'm using OpenSG with Visual C++ with the Intel compiler. The project is > > just a test to see if we can use MFC and OpenSG together. The project > > options for our project are: > > > > /nologo /MD /W3 /GR /GX /O2 /D "WIN32" /D "NDEBUG" /D "_WINDOWS" /D > > "_AFXDLL" /D "OSG_WITH_WIN32" /D WINVER=0x0400 /D _WIN32_WINDOWS=0x0410 /D > > "OSG_BUILD_DLL" /D "OSG_NEW_DLLS" /D "_OSG_HAVE_CONFIGURED_H_" > > /Fp"Release/mfctest.pch" /YX /Fo"Release/" /Fd"Release/" /FD /c > > > > With these options we got off the linker errors. But the executable is very > > unstable. For example we had to change project configuration from "Debug" to > > "Release". With "Debug" we have problem with dynamic linking the libs. When > > executing we get the following message: > > > > First-chance exception in mfctest.exe (OSGSYSTEM.DLL): 0xC00000005: Access > > Violation. > > > > So we had to change to the "Release" configuration. I have no idea why! > > > > By the way when we transpose the following code lines then the executable is > > crashing: > > > > mgr->setWindow(window); > > mgr->setRoot(scene); > > > > regards, > > Christian > > > > > > ----- Original Message ----- > > From: "Gerrit Voss" <vo...@ca...> > > To: "Christian Henning" <he...@ro...> > > Cc: <ope...@li...> > > Sent: Thursday, March 07, 2002 4:23 AM > > Subject: Re: [Opensg-users] strange linker error > > > > > > > >>Hi, > >> > >>could you send me the commandline you use to compile mfctestView.obj. > >>I'll guess you miss some of the flags/defines (e.g. -DOSG_BUILD_DLL > >>-DOSG_NEW_DLLS) because your missing symbols should read > >> > >>__imp_?NullFC@osg@@3VNullFieldContainerPtr@1@B > >>__imp_?_type@WIN32Window@osg@@0VFieldContainerType@2@A > >> > >> From the later one I guess you are talking about the 1.0 version > >>of OpenSG ??. > >> > >>In general and for all, please use the flags provided by > >> > >>osg-config --cflags > >>osg-config Win32 --lflags > >>osg-config Win32 --llib > >> > >>especialy on windows, many symbols are available from different sources > >>and if you catch the wrong ones you program dumps core wherever you > >>would not expect it to do ;-). Unfortunately the 1.0 version of > >>osg-config --llibs is broken, but you can cut&paste all of it and > >>replace $win_sys_libs with opengl32.lib glu32.lib gdi32.lib. Or update > >>your version of OpenSG ;-) > >> > >>Christian Henning wrote: > >> > >> > >>>Hi there, > >>> > >>>at this moment I try to link my MFC project, but I get strange linker > >>> > > error. > > > >>>Basically I have unresolved symbols. I include all libs of your lib > >>>directory. I can even found the symbols in the libs. Here are the > >>> > > errors: > > > >> > >>Did you find these Symbols within the dynamic link (.dll) or the import > >>(.lib) library, if within the dll it is irrelevant as only the import > >>lib is used at link time. > >> > >> > >>>mfctestView.obj : error LNK2001: unresolved external symbol "private: > >>> > > static > > > >>>class osg::FieldContainerType osg::WIN32Window::_type" > >>>(?_type@WIN32Window@osg@@0VFieldContainerType@2@A) > >>>mfctestView.obj : error LNK2001: unresolved external symbol "class > >>>osg::NullFieldContainerPtr const osg::NullFC" > >>>(?NullFC@osg@@3VNullFieldContainerPtr@1@B) > >>> > >>> > >>regards, > >> Gerrit > >> > >> > >> > > > > > > _______________________________________________ > > Opensg-users mailing list > > Ope...@li... > > https://lists.sourceforge.net/lists/listinfo/opensg-users > > > > > > _______________________________________________ > Opensg-users mailing list > Ope...@li... > https://lists.sourceforge.net/lists/listinfo/opensg-users > > |
From: Erik F. <fe...@st...> - 2002-03-08 02:52:42
|
Hi, I'm trying to create some Hotkeys in my Application to apply preset views. Like CTRL-1 is a front view, CTRL-2 a side view from the left , CTRL-3 from the right and so on... Nothing really tricky, but I've got problems finding the methods to do this. Basically I just need to set a position from where I'm looking and a position to where I'm looking. I can get the OSGNavigator from my SimpleSceneManager, but I can't access any methods to do this. Here's what I've tried: Since the FlyNavigator is private, I can't access its methods 'setFrom' and 'setAt', which would be exactly what I needed. I can get the matrix of the FlyNavigator but changing it doesn't have any effect. The OSGNavigator lets me set the center and the distance, but I still need to adjust the rotation, which again I can't access. Will there be some corresponding methods in OSGNavigator in the near future, or is there another way ? Thanx !! Erik |
From: Gerrit V. <vo...@ca...> - 2002-03-08 01:36:46
|
Hi, could you send me the complete build log (compile + link option) so I can have a closer look at it. Thanks Gerrit Christian Henning wrote: > Thanks for answer. > > I'm using OpenSG with Visual C++ with the Intel compiler. The project is > just a test to see if we can use MFC and OpenSG together. The project > options for our project are: > > /nologo /MD /W3 /GR /GX /O2 /D "WIN32" /D "NDEBUG" /D "_WINDOWS" /D > "_AFXDLL" /D "OSG_WITH_WIN32" /D WINVER=0x0400 /D _WIN32_WINDOWS=0x0410 /D > "OSG_BUILD_DLL" /D "OSG_NEW_DLLS" /D "_OSG_HAVE_CONFIGURED_H_" > /Fp"Release/mfctest.pch" /YX /Fo"Release/" /Fd"Release/" /FD /c > > With these options we got off the linker errors. But the executable is very > unstable. For example we had to change project configuration from "Debug" to > "Release". With "Debug" we have problem with dynamic linking the libs. When > executing we get the following message: > > First-chance exception in mfctest.exe (OSGSYSTEM.DLL): 0xC00000005: Access > Violation. > > So we had to change to the "Release" configuration. I have no idea why! > > By the way when we transpose the following code lines then the executable is > crashing: > > mgr->setWindow(window); > mgr->setRoot(scene); > > regards, > Christian > > > ----- Original Message ----- > From: "Gerrit Voss" <vo...@ca...> > To: "Christian Henning" <he...@ro...> > Cc: <ope...@li...> > Sent: Thursday, March 07, 2002 4:23 AM > Subject: Re: [Opensg-users] strange linker error > > > >>Hi, >> >>could you send me the commandline you use to compile mfctestView.obj. >>I'll guess you miss some of the flags/defines (e.g. -DOSG_BUILD_DLL >>-DOSG_NEW_DLLS) because your missing symbols should read >> >>__imp_?NullFC@osg@@3VNullFieldContainerPtr@1@B >>__imp_?_type@WIN32Window@osg@@0VFieldContainerType@2@A >> >> From the later one I guess you are talking about the 1.0 version >>of OpenSG ??. >> >>In general and for all, please use the flags provided by >> >>osg-config --cflags >>osg-config Win32 --lflags >>osg-config Win32 --llib >> >>especialy on windows, many symbols are available from different sources >>and if you catch the wrong ones you program dumps core wherever you >>would not expect it to do ;-). Unfortunately the 1.0 version of >>osg-config --llibs is broken, but you can cut&paste all of it and >>replace $win_sys_libs with opengl32.lib glu32.lib gdi32.lib. Or update >>your version of OpenSG ;-) >> >>Christian Henning wrote: >> >> >>>Hi there, >>> >>>at this moment I try to link my MFC project, but I get strange linker >>> > error. > >>>Basically I have unresolved symbols. I include all libs of your lib >>>directory. I can even found the symbols in the libs. Here are the >>> > errors: > >> >>Did you find these Symbols within the dynamic link (.dll) or the import >>(.lib) library, if within the dll it is irrelevant as only the import >>lib is used at link time. >> >> >>>mfctestView.obj : error LNK2001: unresolved external symbol "private: >>> > static > >>>class osg::FieldContainerType osg::WIN32Window::_type" >>>(?_type@WIN32Window@osg@@0VFieldContainerType@2@A) >>>mfctestView.obj : error LNK2001: unresolved external symbol "class >>>osg::NullFieldContainerPtr const osg::NullFC" >>>(?NullFC@osg@@3VNullFieldContainerPtr@1@B) >>> >>> >>regards, >> Gerrit >> >> >> > > > _______________________________________________ > Opensg-users mailing list > Ope...@li... > https://lists.sourceforge.net/lists/listinfo/opensg-users > |
From: Dirk R. <re...@ig...> - 2002-03-07 18:55:25
|
On Mar 7, 7:45pm, Christian Henning wrote: > Subject: Re: [Opensg-users] strange linker error > Thanks for answer. > > I'm using OpenSG with Visual C++ with the Intel compiler. The project is > just a test to see if we can use MFC and OpenSG together. In general that's possible, we're doing it. > The project > options for our project are: > > /nologo /MD /W3 /GR /GX /O2 /D "WIN32" /D "NDEBUG" /D "_WINDOWS" /D > "_AFXDLL" /D "OSG_WITH_WIN32" /D WINVER=0x0400 /D _WIN32_WINDOWS=0x0410 /D > "OSG_BUILD_DLL" /D "OSG_NEW_DLLS" /D "_OSG_HAVE_CONFIGURED_H_" > /Fp"Release/mfctest.pch" /YX /Fo"Release/" /Fd"Release/" /FD /c > > With these options we got off the linker errors. But the executable is very > unstable. For example we had to change project configuration from "Debug" to > "Release". With "Debug" we have problem with dynamic linking the libs. When > executing we get the following message: > > First-chance exception in mfctest.exe (OSGSYSTEM.DLL): 0xC00000005: Access > Violation. > > So we had to change to the "Release" configuration. I have no idea why! One thing you always have to watch out for with Windows is using the correct run-time lib when creating the DLL and the app. They absolutely have to use the same /MD or /MDd setting, otherwise hell breaks loose. If that's not the problem I don't know what could be wrong. MFC probably has other additional constraints. Gerrit? > By the way when we transpose the following code lines then the executable is > crashing: > > mgr->setWindow(window); > mgr->setRoot(scene); Well, that's the reason they're in that order. :) setWindow should be the first call made to the SceneManager, because other calls might depend on the window being set (to allow the default initialisation that the SM does). >-- End of excerpt from Christian Henning Hope it helps Dirk -- -- -- Dirk Reiners OpenSG Forum di...@op... -- The OpenSG Open Source Scenegraph: http://www.opensg.org -- Announcements: http://lists.sf.net/lists/listinfo/opensg-announce |
From: Christian H. <he...@ro...> - 2002-03-07 18:46:32
|
Thanks for answer. I'm using OpenSG with Visual C++ with the Intel compiler. The project is just a test to see if we can use MFC and OpenSG together. The project options for our project are: /nologo /MD /W3 /GR /GX /O2 /D "WIN32" /D "NDEBUG" /D "_WINDOWS" /D "_AFXDLL" /D "OSG_WITH_WIN32" /D WINVER=0x0400 /D _WIN32_WINDOWS=0x0410 /D "OSG_BUILD_DLL" /D "OSG_NEW_DLLS" /D "_OSG_HAVE_CONFIGURED_H_" /Fp"Release/mfctest.pch" /YX /Fo"Release/" /Fd"Release/" /FD /c With these options we got off the linker errors. But the executable is very unstable. For example we had to change project configuration from "Debug" to "Release". With "Debug" we have problem with dynamic linking the libs. When executing we get the following message: First-chance exception in mfctest.exe (OSGSYSTEM.DLL): 0xC00000005: Access Violation. So we had to change to the "Release" configuration. I have no idea why! By the way when we transpose the following code lines then the executable is crashing: mgr->setWindow(window); mgr->setRoot(scene); regards, Christian ----- Original Message ----- From: "Gerrit Voss" <vo...@ca...> To: "Christian Henning" <he...@ro...> Cc: <ope...@li...> Sent: Thursday, March 07, 2002 4:23 AM Subject: Re: [Opensg-users] strange linker error > > Hi, > > could you send me the commandline you use to compile mfctestView.obj. > I'll guess you miss some of the flags/defines (e.g. -DOSG_BUILD_DLL > -DOSG_NEW_DLLS) because your missing symbols should read > > __imp_?NullFC@osg@@3VNullFieldContainerPtr@1@B > __imp_?_type@WIN32Window@osg@@0VFieldContainerType@2@A > > From the later one I guess you are talking about the 1.0 version > of OpenSG ??. > > In general and for all, please use the flags provided by > > osg-config --cflags > osg-config Win32 --lflags > osg-config Win32 --llib > > especialy on windows, many symbols are available from different sources > and if you catch the wrong ones you program dumps core wherever you > would not expect it to do ;-). Unfortunately the 1.0 version of > osg-config --llibs is broken, but you can cut&paste all of it and > replace $win_sys_libs with opengl32.lib glu32.lib gdi32.lib. Or update > your version of OpenSG ;-) > > Christian Henning wrote: > > > Hi there, > > > > at this moment I try to link my MFC project, but I get strange linker error. > > Basically I have unresolved symbols. I include all libs of your lib > > directory. I can even found the symbols in the libs. Here are the errors: > > > Did you find these Symbols within the dynamic link (.dll) or the import > (.lib) library, if within the dll it is irrelevant as only the import > lib is used at link time. > > > > > mfctestView.obj : error LNK2001: unresolved external symbol "private: static > > class osg::FieldContainerType osg::WIN32Window::_type" > > (?_type@WIN32Window@osg@@0VFieldContainerType@2@A) > > mfctestView.obj : error LNK2001: unresolved external symbol "class > > osg::NullFieldContainerPtr const osg::NullFC" > > (?NullFC@osg@@3VNullFieldContainerPtr@1@B) > > > > regards, > Gerrit > > |
From: Gerrit V. <vo...@ca...> - 2002-03-07 03:25:40
|
Hi, could you send me the commandline you use to compile mfctestView.obj. I'll guess you miss some of the flags/defines (e.g. -DOSG_BUILD_DLL -DOSG_NEW_DLLS) because your missing symbols should read __imp_?NullFC@osg@@3VNullFieldContainerPtr@1@B __imp_?_type@WIN32Window@osg@@0VFieldContainerType@2@A From the later one I guess you are talking about the 1.0 version of OpenSG ??. In general and for all, please use the flags provided by osg-config --cflags osg-config Win32 --lflags osg-config Win32 --llib especialy on windows, many symbols are available from different sources and if you catch the wrong ones you program dumps core wherever you would not expect it to do ;-). Unfortunately the 1.0 version of osg-config --llibs is broken, but you can cut&paste all of it and replace $win_sys_libs with opengl32.lib glu32.lib gdi32.lib. Or update your version of OpenSG ;-) Christian Henning wrote: > Hi there, > > at this moment I try to link my MFC project, but I get strange linker error. > Basically I have unresolved symbols. I include all libs of your lib > directory. I can even found the symbols in the libs. Here are the errors: Did you find these Symbols within the dynamic link (.dll) or the import (.lib) library, if within the dll it is irrelevant as only the import lib is used at link time. > > mfctestView.obj : error LNK2001: unresolved external symbol "private: static > class osg::FieldContainerType osg::WIN32Window::_type" > (?_type@WIN32Window@osg@@0VFieldContainerType@2@A) > mfctestView.obj : error LNK2001: unresolved external symbol "class > osg::NullFieldContainerPtr const osg::NullFC" > (?NullFC@osg@@3VNullFieldContainerPtr@1@B) > regards, Gerrit |
From: Christian H. <he...@ro...> - 2002-03-06 14:45:49
|
Hi there, at this moment I try to link my MFC project, but I get strange linker error. Basically I have unresolved symbols. I include all libs of your lib directory. I can even found the symbols in the libs. Here are the errors: mfctestView.obj : error LNK2001: unresolved external symbol "private: static class osg::FieldContainerType osg::WIN32Window::_type" (?_type@WIN32Window@osg@@0VFieldContainerType@2@A) mfctestView.obj : error LNK2001: unresolved external symbol "class osg::NullFieldContainerPtr const osg::NullFC" (?NullFC@osg@@3VNullFieldContainerPtr@1@B) I use the Intel Compiler. Regards, Christian |