Re: compiling brat/brop/brag in RH7.1 pii computers

From: Kris Hagel (hagel@CyclotronMail.tamu.edu)
Date: Wed Oct 17 2001 - 16:11:53 EDT

  • Next message: Christian Holm Christensen: "Old stuff in /afs/rhic/opt/brahms (Was Re: compiling brat/brop/brag in RH7.1 pii computers)"

    Christian,
    The reason I continue to use the root in /afs/rhic/opt/brahms/root (root 3.02 
    now) is because I almost exclusively use my own brat in my own directory. Since
    the root and brat files are all mixed together in/afs/rhic/opt/brahms/pro/lib 
    as well /afs/rhic/opt/brahms/pro/bin, I find it an extra level of confusion 
    when linking things if I want to link to my personal brat and then send link 
    paths into a directory (afs/rhic/opt/brahms/pro/lib) which has a mixture of both
    sets of libraries (ROOT + BRAT).  Also, when doing that, I have no idea which
    version of root I might be linking to.  And there is always the issue that when
    I have something working, someone else will upgrade root and replace all the
    root libraries in those directories and my stuff (or someone's stuff who I am
    helping) will cease to work with no explanation whatsoever.  Having root in its
    own separate directory which will not be messed with once it is built completely
    sidesteps that problem and keeps things working.
    
    SO, DO NOT DELETE the stuff in /afs/rhic/opt/brahms/root (except versions 
    earlier than root 3.02) until you have satisfied my concerns noted above.
    
    Regards
    
    Kris
    
    
       Quoting Christian Holm Christensen <cholm@hehi03.nbi.dk>:
    
       > Hi Hiro, 
       > 
       > On Wed, 17 Oct 2001 05:58:59 -0500 (CDT)
       > Hironori Ito <hito@students.phsx.ukans.edu> wrote
       > concerning "compiling brat/brop/brag in RH7.1 pii computers":
       > >     Hello.  I finally checked if I can compile brat/brop/brag on
       > > RH7.1 pii machines.  And, the answer is "yes".  Here is what
       it can
       > be
       > > done.
       > > 
       > > 1.  BRAT/BROP
       > >   CC=egcs CXX=egcs++ ./configure
       > > 
       > > This should creates the binary programs which runs on RH6.2 or RH7.1
       > > 
       > > 2.  BRAG
       > >   CC=egcs  ./configure --with-cern-libdir=/home/cern/2000/lib
       > > --with-cern-bindir=/home/cern/2000/bin
       > 
       > Just like above, but since Red Hat (sigh) does not ship a egcs77, you
       > need to use /usr/bin/i386-glibc21-linux-g77.  That is 
       > 
       >   F77=/usr/bin/i386-glibc21-linux-g77 ./configure ... 
       > 
       > > This will create the binary program which sadly only runs on RH7.1 .
       > (Any
       > > suggestions???)  But, this should not be a big problem since BRAG can
       > be 
       > > compiled within a couple minutes unlike BRAT, which takes forever.  
       > > Therefore, just compile brag if it does not work. 
       > 
       > One of the reasons why BRAT takes a long time (ca. 1/2 hour) is that
       > we use optimisation now, which wasn't done in BRAT 1.  This can be
       > turned off with configure option --disable-optimization. 
       > 
       > Another reason is the automatic dependency generation done by
       > automake.  This is so that make can figure out which files to
       > recompile in case of changes to BRAT/ROOT/system headers - i.e., a
       > Good Thing.  This can be disables with by putting no-dependencies in
       > the AUTOMAKE_OPTIONS variable of the toplevel Makefile.am - NOTE: this
       > is not recommended! 
       > 
       > Thirdly, there are a number of redundant and/or unused classes in BRAT
       > that prolongs the build time considerably, not least because of the
       > two above points. 
       > > 
       > >     Now, I do not know the situations for the other programs out there
       > > (brahvo, badtrip, tpc3d, etc...), but since all programs are based on
       > > root (and brat), I assume that they should work.
       > 
       > They should be configured like done above, that is, with 
       > 
       >   CC=egcs CXX=egcs++ F77=/usr/bin/i386-glibc21-linux-g77 ./configure
       > ...
       > 
       > >     Hiro
       > > 
       > > NOTE:  I was not able to compile brat without using the above method
       > > on the RH7.1 pii machines.  Probably, I am using wrong root for
       > > RH7.1 since I have no trouble compiling at home using RH7.1. 
       > 
       > If you have a ROOT that's compiled with 'Red Hat GCC 2.96', and you
       > try to link BRAT build with GCC 2.91 against these libraries, you may
       > find some trouble.  In general one should try to compile all stuff
       > (especially C++ code) with the same compiler.   That said, it's
       > usually not a problem, but since ROOT uses many features (streches
       > into almost every corner of 'phase-space') you may see some problems. 
       > 
       > BTW, do you know if there are RPMs of GCC 2.95.3 out there?  If so, it
       > may be a good idea to install those too, since 2.95.3 is rather stable
       > compiler - and in fact never than 'GCC 2.96'. 
       > 
       > Just out of intrest, are you using the RPMs I made some time ago? 
       > 
       > BTW, I jsut remembered we have some old ROOT installs lying around on
       > /afs/rhic/opt/brahms.  I'll remove these, since ROOT 3.02 is now in
       > the production tree, and these old ROOTs should not be used (see one
       > of my answers to Hiro some weeks ago). If you object to this, SCREAM
       > now, and tell me your reasons - they better be good, REALLY GOOD,
       > otherwise I'm going to delete the stuff tomorrow (CET). 
       > 
       > Yours, 
       > 
       > Christian Holm Christensen -------------------------------------------
       > Address: Sankt Hansgade 23, 1. th.           Phone:  (+45) 35 35 96 91 
       >          DK-2200 Copenhagen N                Cell:   (+45) 28 82 16 23
       >          Denmark                             Office: (+45) 353  25 305 
       > Email:   cholm@nbi.dk                        Web:    www.nbi.dk/~cholm
       > 
    



    This archive was generated by hypermail 2b30 : Wed Oct 17 2001 - 16:12:32 EDT