divVerent wrote:Also, I really doubt the lighting preview code of that helps AT ALL, as it is also in GtkRadiant 1.5. It can only be used in Doom 3 mode. The real thing to do would be making it support Quake 3 mode too. And I have no clue about that part (or even why it does nothing in Quake 3 maps at all).

It works in Quake 3(Version 1) mode. It is based off of his QRazor-FX engine.

Each mode is a game in the list, Version 1 is Quake 3, Version 2 is Doom 3, and Version 3 is Quake 4.

Which version? xrealradiant based on darkradiant or gtkradiant previously known as xrealradiant? I tried to compile latter and got ton of errrors.

Ok, found this. I'll try to compile later back at home.
http://xreal.sourceforge.net/xrealwiki/SourceCode

Just get the xreal svn, compile with scons mapping=yes, and if you get errors its because you are missing a dependency. I am not sure if the original GtkRadiant fork for XreaL is still in SVN, though.

Yes, it is.

I don't think NetRadiant works with the latest NVIDIA drivers(180.22). I get this when NetRadiant loads(as well as with the original GtkRadiant 1.5.0):

Code: Select all

(radiant.x86:24311): GdkGLExt-CRITICAL **: gdk_gl_window_impl_x11_make_context_current: assertion `GDK_IS_GL_CONTEXT_IMPL_X11 (glcontext)' failed

Segmentation fault

This probably gets fixed by reinstalling the nvidia driver and a reboot, as it only happens when the GL context creation failed.

And that appears to be what happened!

Thanks.

Can anyone try to compile the current source on OS X and run it in X11?

Make sure to use fink and have the following packages installed:

atk1 atk1-shlibs gettext gettext-bin gettext-dev glib glib-shlibs glib2 glib2-dev glib2-shlibs gtk+2 gtk+2-dev gtk+2-shlibs gtkglext1 gtkglext1-shlibs libgettext3-shlibs libiconv libiconv-bin libiconv-dev libxml2 libxml2-bin libxml2-shlibs make pango1-dev pango1-shlibs pango1-xft2 pango1-xft2-dev pango1-xft2-shlibs pkgconfig svn

Also, it has to be run in X11.

Hi,
I'm a mapper for another q3 engine based game and I have tried NetRadiant few days ago. Thanks a lot for this stabilized radiant 1.5 (now I will eventually map under linux ).

I have found a bug, when using the "regroup" feature, it leaves entity with no brush (and Q3 don't want to load a map if it found an entity with no brush).

Basicly, if you try to regroup all brushes of an entity into another, the first entity is not deleted.

(BTW I am using the bleeding edge svn version )

I cannot reproduce this problem. Did you compile your build, or take the Windows build download?

Before: http://emptyset.endoftheinternet.org/~r ... 091db6.png
After: http://emptyset.endoftheinternet.org/~r ... 8258ac.png

Did you do something different from this? Would need that info so I can reproduce the problem.

I build it last week (linux Debian): NetRadiant 1.5.0-svn169
Unfortunately, I can't reproduce it at will (just tried to start a new map and try lot of combination without success).

But with a whole map, it happened to me often (around 10-15 times after one week I use NetRadiant).
Sometime I got segfault just after clicking on "regroup".
On one of my testbox it happened everytime I forgot to hit Ctrl-Alt E (selecting the whole group) before trying to regroup.
(on a new map, it used to move brushes into worldspawn when I forget to hit ctrl-alt E).

Sorry, I can't help you more...

Yes, moving them into worldspawn is intended if no entity but just brushes are selected. That way, you can remove single brushes from an entity.

Looks like I'll need a gdb trace on that problem then. Until then, regularily save your map.

divVerent wrote:Can anyone try to compile the current source on OS X and run it in X11?

Make sure to use fink and have the following packages installed ...

At least with my PowerPC machine, NetRadiant (1.5.0svn168) seems to build on OS X without any problems. I've seen only one serious issue with usability: attempting to place a MD3 model in a map or opening a map with a model already in it will crash the application.

I had some misgivings about Fink in the past, so I've long since switched to MacPorts. I'd be willing to build again with the Fink libraries just to see if it makes a difference.

Too bad, I don't have a Mac so I can't help you. It works on the Mac I can remotely log in to, but very slowly.

Can you catch that bug in gdb?

divVerent wrote:Too bad, I don't have a Mac so I can't help you. It works on the Mac I can remotely log in to, but very slowly.

Can you catch that bug in gdb?

I can certainly try. Now we're getting a little beyond my level of expertise. I can get as far as doing backtrace before I completely have no idea what I'm doing, sorry. I'll email you what crash logs I have.

airscout wrote:... At least with my PowerPC machine, NetRadiant (1.5.0svn168) seems to build on OS X without any problems....

... to build again with the Fink libraries just to see if it makes a difference.

I should note that that was a typo. I've in fact been working with 1.5.0svn169.

Also I did attempt to build with the Fink tools and libraries instead of those from Macports, with even worse results. It could never finish compiling, always stopping with errors.

heres something i whipped up in blender

take it or leave it, tell me what you think

divVerent wrote:Can anyone try to compile the current source on OS X and run it in X11?

Make sure to use fink and have the following packages installed:

atk1 atk1-shlibs gettext gettext-bin gettext-dev glib glib-shlibs glib2 glib2-dev glib2-shlibs gtk+2 gtk+2-dev gtk+2-shlibs gtkglext1 gtkglext1-shlibs libgettext3-shlibs libiconv libiconv-bin libiconv-dev libxml2 libxml2-bin libxml2-shlibs make pango1-dev pango1-shlibs pango1-xft2 pango1-xft2-dev pango1-xft2-shlibs pkgconfig svn

Also, it has to be run in X11.

airshot wrote: At least with my PowerPC machine, NetRadiant (1.5.0svn168) seems to build on OS X without any problems. I've seen only one serious issue with usability: attempting to place a MD3 model in a map or opening a map with a model already in it will crash the application.

Argh, I just noticed this thread, as I'm not at all a modeler. I'll give it a shot on my PPC and Intel Macs to see what I can produce, too.

fabz0r wrote:heres something i whipped up in blender

take it or leave it, tell me what you think

Nice.

BTW, would there be a way to run graphical applications for testing this on your build machine? I have no idea if there is something similar to remote desktop for OS X. X11 forwarding does work, but is no option, as it is awfully slow and applications often misdraw OpenGL stuff when using it (like, it takes some minutes to start up radiant, and the whole 2D view is awfully blurred - might be a problem between nvidia linux driver and OS X OpenGL libraries).

the last one was i think good for a chunky little copper icon, but i thought the letters of net were a little hard to make out, so i made this second version for the splash screen and the about icon, i hope you like it

I think the main idea of the logo gets lost: it casts the right shadows in all directions. A seemingly impossible object, but very buildable in radiant.

divVerent wrote:BTW, would there be a way to run graphical applications for testing this on your build machine? I have no idea if there is something similar to remote desktop for OS X. X11 forwarding does work, but is no option, as it is awfully slow and applications often misdraw OpenGL stuff when using it (like, it takes some minutes to start up radiant, and the whole 2D view is awfully blurred - might be a problem between nvidia linux driver and OS X OpenGL libraries).

Sure, I can see what I can whip up. It could very well be an incompatibility -- I'll give X forwarding a try from Mac to Mac and see what I get out of it, as well as Debian to Mac. I'll also go about updating xquartz, but the only other option I know of is VNC. Also, Tiger might work a bit better in terms of X forwarding.

Apple's own Remote Desktop app is built upon VNC, but I'll just use the generic vncserver. I'll get that port forwarded tonight (5900), and see how well my connection handles the bandwidth, otherwise it should be able to be tunneled over the ssh connection, too.

As for a test case, since I'm somewhat clueless about importing an MD3 model -- are there any items in the svn repo that I can import that would likely reproduce this, or should I just try to import any old md3?

I can't seem to make this on ubuntu 8.10 64bit,

Checking for GL/gl.h (mesa-common-dev (or another OpenGL library))... not found, please install it or set PKG_CONFIG_PATH right!
To see the failed commands, set DEPENDENCIES_CHECK=verbose
To proceed anyway, set DEPENDENCIES_CHECK=off
make: *** [dependencies-check] Error 1

So I,

apt-get install mesa-common-dev
Reading package lists... Done
Building dependency tree
Reading state information... Done
mesa-common-dev is already the newest version.
0 upgraded, 0 newly installed, 0 to remove and 1 not upgraded.

Not sure what to do here.

airscout wrote:

divVerent wrote:Can anyone try to compile the current source on OS X and run it in X11?

Make sure to use fink and have the following packages installed ...

At least with my PowerPC machine, NetRadiant (1.5.0svn168) seems to build on OS X without any problems. I've seen only one serious issue with usability: attempting to place a MD3 model in a map or opening a map with a model already in it will crash the application.

My main test machine is a MacBook Pro, but I did manage to get it to crash on both a PPC and Intel Mac by importing a md3 model, as well as loading a map (aggressor.map, and turrets_test.map were the two I used). I know that "final_rage-patches.map" worked. So, Mac OS X kindly gives a reasonable backtrace in its crash logs when it has the source available to it. Here's my crash log. Here's a brief gdb session producing the same bt. I can start setting breakpoints and step through, if you'd like.

Anyroad, I installed the dependencies thusly:

Code: Select all

# fink -k install atk1 atk1-shlibs gettext gettext-bin gettext-dev glib glib-shlibs glib2 glib2-dev glib2-shlibs gtk+2 gtk+2-dev gtk+2-shlibs gtkglext1 gtkglext1-shlibs libgettext3-shlibs libiconv libiconv-bin libiconv-dev libxml2 libxml2-bin libxml2-shlibs make pango1-dev pango1-shlibs pango1-xft2 pango1-xft2-dev pango1-xft2-shlibs pkgconfig svn

If these have already been installed then you'll wanna rebuild. There might be a parameter to just grab and cache the source, but I couldn't find it.

Code: Select all

# fink -k rebuild atk1 atk1-shlibs gettext gettext-bin gettext-dev glib glib-shlibs glib2 glib2-dev glib2-shlibs gtk+2 gtk+2-dev gtk+2-shlibs gtkglext1 gtkglext1-shlibs libgettext3-shlibs libiconv libiconv-bin libiconv-dev libxml2 libxml2-bin libxml2-shlibs make pango1-dev pango1-shlibs pango1-xft2 pango1-xft2-dev pango1-xft2-shlibs pkgconfig svn

Also, quick tip, div: 10.5 has gotten rid of the /usr/bin/open-x11 script. It's automatically launched from within launchd, so just using:

Code: Select all

$MY_DIRECTORY/Contents/MacOS/install/radiant.ppc

Works on 10.5. Your other method works just fine in 10.4, but it won't work in 10.5. Apple completely borked up X11, and I'm sure it's at least partially due to the CodeSigning crap they instituted. Anyroad, I'll pm you know when I have VNC setup on the build MacBook, div.

AUR pkgbuild by epsy: http://aur.archlinux.org/packages.php?ID=23523

* New build based on SVN rev 177
http://ingar.satgnu.net/gtkradiant/index.html#download

* Small patch to use firefox as default help browser on UNIX (Who uses netscape anyway these days).
netradiant-177-firefox.patch
Is there a way to use the default browser, whatever that might be?

Okay... so I solved my problem above by removing mesa-common-dev and installing: nvidia-glx-173-dev

nvidia-glx-173-dev also has a GL/gl.h -- this one works on my 64 bit intrepid.

Lixi wrote:Also, quick tip, div: 10.5 has gotten rid of the /usr/bin/open-x11 script. It's automatically launched from within launchd, so just using:

Code: Select all

$MY_DIRECTORY/Contents/MacOS/install/radiant.ppc

Works on 10.5. Your other method works just fine in 10.4, but it won't work in 10.5.

I gave some thought to that issue. NetRadiant.app/Contents/MacOS/netradiant.sh could be modified to something like this:

Code: Select all

#!/bin/bash

# Determine where we are
MY_DIRECTORY="${0%/*}" # cut off the script name
cd "$MY_DIRECTORY/install"

# 10.4 uses Darwin 8.x.x. 10.5 uses 9.x.x
OSrelease=$(uname -r | sed 's,\([0-9]*\)\(.*\),\1,g') # cut off sub versions
OSlimit=9

if [ $OSrelease -lt $OSlimit ] # Is our version less than 9?
then
   # Run on 10.4... and below?
   exec /usr/bin/open-x11 ./radiant.ppc
else
   # Run on 10.5
   # exec ./radiant.ppc # This works, but seems to hang the launcher application on 10.5.    
   screen -d -m -S netradiant ./radiant.ppc # Using screen is probably more flexible anyway.
fi

Although someone should test it on 10.4, Lixi? Perhaps they could even check to see if the screen method can be used in conjunction with open-x11.

Also, I don't know whether it was intentional or not, but the OS X executable has the Nexuiz icon, not NetRadiant's– which might not send the right message.
If needed, here's a replacement for NetRadiant.icns with the "correct" icon.

Will work on that later.

As for the icon - I simply have no application to make .icns files. Thanks for the file.

Running this in screen, however, is a bad idea, as it interferes with the user using screen.

Can you try ./radiant.ppc & instead?

divVerent wrote:As for the icon - I simply have no application to make .icns files. Thanks for the file.

NP. Unfortunately, I just checked out 184 and that .icns file appears to have gotten corrupted somewhere along the way. I'll email you a replacement and put a zipped version here.

divVerent wrote:Can you try ./radiant.ppc & instead?

That works perfectly. Cool.