Hi,
I tried it once again. Still, I cannot build the GUI. First, for some
reason, the "moc"-tool is searched for in /usr/bin, but I do not have it
there. I used a symlink, but it is quite silly as moc is actually in the
PATH. Than, I got errors building the GUI that I have attached.
CU Lars.
error.log.gz (3.15 KB)
Hi Lars,
sorry, I took some days off and I don't have access to my machines right now to have a closer look at the problem, so unfortunately you'll have to wait until next week for an answer..
greetz
Carsten
No problem, the command line is working. I did not find any
documentation about using the pmap stuff, there is no mkpmap in
Radzilla, right? How do I control it (without GUI), I guess there are
some rzpict parameters to use?
CU Lars.
Hi Lars,
I had a look at your error log and it reminded me of something.. it seems that QT still has not decided upon a unique scheme for putting all its necessary stuff, and/or the various OS builders add some further confusion on their own behalf.
According to your log output, the header files are simply not found. If you run the installation script, chose the option number 2
when the script asks you about editing compilation parameters
Do you want to (or need to ..) modify compilation parameters (experts only) ?:
1) Edit basic compilation parameters (compiler options etc.)
2) Edit QT specific parameters (Location of QT libraries etc.)
......
.......
and replace the specific macros with the paths on your system. I remember that for example adding additional include directories
like -I$(QTDIR)/include/qt4/QtCore or sth. like that to the QTINCL line once did the trick.
If you already did run the installation script once and have all commandline stuff built, there is a shortcut for the QT stuff
without running the installation script again and again:
edit directly the file zmake/qtspecial2, and run 'make all' from the qt subdirectory of the radzilla source tree..
I remember that Andreas Noback succeded in compiling the QT interface once, although the unfortunately
the interface didn't work porperly, we didn't find outr back then due to the usual lack of time, I remember that I sent him some ecperimental fix and never came to know if this had an effect or not.
The whole stuff is weird, 'cause I simply use the QT functions like they explain it in their manual, and rely on their boastful promise of platform independence, and - well you see the results.
Alles nur Verarsche 
greetz
Carsten
Hi Carsten!
Thanks for the quick understanding of the "error log"... than the
solution was more than simple, the QTDIR was set to /usr in the config
file. I wonder if I typed that during the installation process, or if
this is somewhere assumed in the Radzilla make system? Because actually,
I have an environment variable QTDIR pointing right to the directory
where the whole stuff is installed.
I will try out now how it feels with Radiance..ahemm...Radzilla and a
GUI 
BTW, do you know Renderpark?
CU Lars.
Ok, before enjoying the GUI, I had to change the files stringlistmodel.C
and fileinfomodel.C in qt/aux a bit. You have a number of "return
(QVariant());" lines in there, which are not accepted in my g++ (3.3
Apple). I changed these, so that a variable is assigned the return value
of QVariant(), and this variable returned than in return(). I guess this
behaviour can be changed with a compiler switch, as it should be allowed
according to the c++-standard one chooses. But I could not find another
way right now.
Still, new errors, new err.out 
CU Lars.
err.out (2.12 KB)
Ok, before enjoying the GUI, I had to change the files stringlistmodel.C
and fileinfomodel.C in qt/aux a bit. You have a number of "return
(QVariant());" lines in there, which are not accepted in my g++ (3.3
hmm, well, this sounds puzzling, the QVariant() stuff should not pose a problem,
its a simple constructor call.
I should note that I did compile everything with gcc version 4.1.0.
Version 3.3 isn't really up to date anymore..
Still, new errors, new err.out
CU Lars.
ok, now there come the usual missing type info messages, each unix flavour makes
different assumptions about what's known or included already in standard library include files
in the specific case, try listing
#include <unistd.h> before <sys/mman.h>, or try adding other usual suspects like
(#include <sys/types.h>
obviuosly in OSX, sys/mman.h assumes the types off_t and caddr_t
to be defined already, but it's not the case with the currently loaded includes in the
Linux case.
Unfortunately, this is a rather simple search and find issue, which however is difficult to do without
a Mac ready at hand...
-cb
Hi Carsten!
hmm, well, this sounds puzzling, the QVariant() stuff should not pose a problem, its a simple constructor call.
As I noted, the call is fine, it is just not allowed INSIDE the return(). I do not know which c++ standard allowed to pass the function call as argument to return().
I'll go and search for the missing include
CU Lars.
Ok Carsten, I have Radzilla 1.2 on Mac OS X 10.3.9 now
After the changes I had mentioned (e.g. moving the constructor calls out
of the return() statement), I had to:
1) add #include <unistd.h> to prev/rzvuwidget.C and rzprevwindow.C
2) modify zmake/qtspecial2 to contain QTLIBS= -F$(QTDIR)/lib -framework
QtGui -framework QtCore
Now the problems are:
1) The fonts are unreadable (small)
2) I cannot make it load an octree (always says "cannot load file..."
CU Lars.