January the 2nd, 2000 The alpha version has been released yesterday. Today I have been busy with moving the project site to the SourceForge. As for now there will be the web site available and I will give it a try to set up the freeglut mailing lists. There will be no CVS access available for now (my dialup internet connection sucks so badly that I wouldn't be able to work on the project). After I am done with that, I will try announcing the project on www.opengl.org. Of other things, there has been rumours floating round the routers and hubs about Mark Kilgard changing the GLUT's license, but this is unconfirmed. It would be really cool if he did so, there's no better motivation to work than a bit of sound competition. As for me, I already put too much work into the freeglut project to terminate it just now. We'll see what happens next. January the 4th, 2000 Ho-ho-ho. Freeglut-1.3 works fine with `Tux the Penguin and the Quest for Herring'. At least that's what Steve Baker, the author, says. Not only that, Steve has provided us with the joystick code (from his great PLIB project) and numerous hints and tips about making freeglut more useful (I will try to put the results of our discussion on the structure page). As for other issues -- I promise I will start the Win32 port this weekend. BTW. -- is there a decent cross compiler for Linux that generates Win32 executables, so that I don't have to use windows for development? And what about Wine OpenGL support? The package is now some 40kB smaller than the previous one. Did some general clean ups, removed unnecessary configure scripts in the genfonts directory, the configure cahce, the Makefiles and so on. Also, I have started introducing the new internal structure, as described here. January the 6th, 2000 The missing glutInit() issue seems to be solved. Chris Purnell (fxGLUT author) says that the GLUT window and menu creation functions call glutInit() if the caller didn't do that before. The enumerations for GLUT_KEY_UP and GLUT_KEY_RIGHT were accidentally swapped. They should be OK now. Hope the rest is OK :) Added two new API calls to freeglut-1.3 -- glutBitmapHeight() and glutStrokeHeight(), as suggested by Steve Baker. This won't break the GLUT compatibility a lot, and I've heard it can be useful. Here you go. The structure plans page has been updated. The numerous feature hints from opengl-gamedev-l readers have been added. Somebody (Chris?) hinted me that the stroke fonts can be found in the XFree86 sources. I browsed through it and -- presto. Now I only need to define the stroke fonts structure (should be very similiar to bitmapped one) and write quite a simple parser. I've spent the (late) evening doing the init display string parsing and making my logics classes homework :) Both is harder than I primarily thought, but fortunately for me I can commit errors in one of those tasks. Guess which one? Argh. :) January the 8th, 2000 First of all, both the missing glutInit() and glutGet(GLUT_WINDOW_[X|Y]) issues are fixed now. The first problem was solved thanks to Chris Purnell, who showed me the way GLUT dealt with the problem. Good for me there's someone who reads it's source code (I just felt it would be unfair for me to do so :D). The second problem was solved by adding a XMoveWindow call just after the window creation and initial mapping. This is strange. Maybe one of the Xlib calls between the creation and mapping spoiled the window's coordinates? This makes even more GLUT tests work fine. The tests can be found in any decent GLUT 3.7 distribution. Following tests produce a FAIL message: test18.c (some layer issues), test19.c (see the GLUT_NORMAL_DAMAGED issue on the progress page), test22.c (window stacking/visibilty problems), test23.c (glutInitDisplayString() is unfinished), test25.c (the freeglut fonts differ a bit from GLUT's), test28.c (-iconic handling is missing(?)). Gee :) I've spent another hour doing the glutInitDisplayString(), but it still is far from being complete. And I've also started gathering information on doing the game mode stuff. The video mode enumeration in both cases will be painful. There is a big issue with the window contents redrawing. Right now, it is forced every time there are no events pending in the main loop. I know it's wrong, but it without such an approach some of the test programs freeze soon after starting. Could someone peer-review the main loop code please? I have decided to start the Win32 port this weekend. As for now, the code compiles under vc++5.0. I will start making it work as soon as I download the pthreads library and the newest version of GLib for Windows. It was quite a good idea to start the port, as the Microsoft's compiler generates much more warnings and I had quite a few things fixed for free. January the 9th, 2000 Doing the Win32 port all the day... Actually, there is enough code to make it work, however I am sure only of that it compiles (more or less). I need to download the pthreads-win32 library to get the GLib working first, and somehow I was unable to do it during the weekend. Once again -- the Win32 port does not work yet. Oh, and I need adding the __declspec(dllexport) thing... After it starts working, I'll have to clean up the code a bit, I guess... January the 10th, 2000 Here I am at three o'clock am, half-awake, uploading and downloading things for freeglut. I never thought I'd be able to force myself getting up that early :) January the 16th, 2000 Both the flu and a terrible feeling about the dialup bills made me slow down a bit, the internet activity I mean :). But here I am again uploading the latest snapshot. The biggest news about it is the Win32 port working, with nearly all features you can find in the X11 version implemented (still, it needs some debugging). For the Unix port, game mode support (loosely based on SDL source code posted at Neal Tringham's www.pseudonymz.demon.co.uk) and numerous bug fixes have been introduced. In order to compile the Win32 version, you'll need pthreads-win32 library (see sourceware.cygnus.org), the GLib-1.2.x (www.gtk.org, I've been using the 1.2.6), a working native compiler (Microsoft VisualC++ 5.0 in my case), and a bit of patience. See the project files I've supplied for some definitions needed (FREEGLUT_LIBRARY_BUILD needs to be declared when building the DLL), and don't forget freeglut joystick code is making use of Win32 multimedia extensions (link against winmm.lib). Be prepared to meet the fact Mesa 3.1 (or at least my compilation) doesn't work very well with this snapshot -- something's messed up when changing WGL contexts. This is really strange, as the Microsoft's OpenGL works pretty fine, as does Dominik Behr's miniGL thing. The assumption is that I've taken some approach that somehow is valid with Microsoft's drivers, but is not OpenGL conformant. Could anyone check this out please? My plan for next week is to add some lesser features missing, and start learning maths as the session at my university is coming in really fast :) This way or another, expect the next release not any sooner than next weekend (given that no nasty bugs get digged out). Argh. Don't be surprised if the code doesn't compile under X-11 other than XFree86.
It could fail when trying to include the X11/extensions/xf86vmode.h include header,
in that case just comment out that inclusion line (found in freeglut_internal.h).
Is there any intelligent way to detect the existence of an include header, and if
it's autoconf to be the answer, how to use it?
|