Hello,
I am almost embarrassed to send this mail to indicate that I solved the
problem and what the simple nature of the problem was. It was my fault. My
problem with having the undefined symbol had to do with the fact that at BNL
on the pii's, I was just running root with all of the graphics. When I got
the e-mail from Yury, in order to avoid the long waits on the orange screen, I
did a root -b. But that does not load libGui.so in the $ROOTSYS/lib
directory. If one does a root -b, one needs to load libGui.so by hand. That
is typically not done, because if you do a root -b, you are not expecting to
do any graphics.
I polked into the .cshrc in the nbi account and the reason Ian had the problem
is that the $ROOTSYS points to an old version of ROOT. You could make this
transparent by not defining ROOTSYS in your .cshrc. If you execute
/home/brahmlib/brahms_login.csh, a ROOTSYS will be defined as well as BRATSYS
which points to the ./pro directory. The ROOTSYS is forseen to point to a
version of ROOT with which BRAT was compiled.
The error seen there (as mine even though the cause was different) indicates
different versions of root or different compilers. In this case, it was the
different version of root. I know I am getting lamblasted for putting the pro
on something relatively new, but this problem would have occurred when the pro
was changed to something else no matter how mature the code. Anyway, I also
agree that code in the pro directory should be at least somewhat more mature
than in the new directory and will make efforts to keep it like that.
The last problem had to do with loading libBrat_Monitor.so That needs to be
loaded sometime between libBrat_Base.so and the first time you load a
libBrat_??? which contains a monitor (ie something that inherits from
BrBaseMonitor) I am giving some thought on how this can be made more
transparent. I have some ideas and will let this list know if it works.
Kris
This archive was generated by hypermail 2b29 : Tue Feb 01 2000 - 20:35:05 EST