[Dune] Verwendung von DUNE und der aktuellen UG-Version

Mon May 21 16:49:36 CEST 2007

I don't know. We disabled it already, what made the standard ug compile
without errors. Does DUNE require GRAPE? If so, have I to get it somewhere
else?

Oliver Sander wrote:
> How did you get -DDebug as a compiler option?  It is not
> supposed to be set.
>
> --
> Oliver
>
> lenzen at scai.fraunhofer.de wrote:
>
>>Dear Oliver,
>>
>>I guessed you would ask for that information. I tried to use standard
>>options on a quite usual system. Several experiments led to different
>>errors, so I'll try to repeat them and give you as much information as
>>possible. The PC I used currently runs under windows, so I'll use a
>> system
>>with - as far as I know - the same specifications. In all cases I gave
>> the
>>option --enable-dune to autogen.sh, which should be passed to configure,
>>but I'm not sure if there is any effect to this option.
>>
>>1. uname -a gives:
>>
>>Linux briggs 2.6.9-11.ELsmp #1 SMP Thu Jun 9 15:33:26 CDT 2005 i686 i686
>>i386 GNU/Linux
>>
>>2. gcc -v (and g++ -v) give:
>>
>>Lese Spezifikationen von /usr/lib/gcc/i386-redhat-linux/3.4.6/specs
>>Konfiguriert mit: ../configure --prefix=/usr --mandir=/usr/share/man
>>--infodir=/usr/share/info --enable-shared --enable-threads=posix
>>--disable-checking --with-system-zlib --enable-__cxa_atexit
>>--disable-libunwind-exceptions --enable-java-awt=gtk
>>--host=i386-redhat-linux
>>Thread-Modell: posix
>>gcc-Version 3.4.6 20060404 (Red Hat 3.4.6-8)
>>
>>3. echo $UGROOT gives:
>>/home/clenzen/dipl_ware/ug
>>
>>4. The "vanilla" ug version compiles with gcc when using autogen.sh and
>>make. Of course the libraries is not accepted by the beta5-version of
>>DUNE.
>>
>>5. Compiling it with g++ (or gcc -xc++, both forced by setting ARCH_CC
>>when calling make) leads to an impressive number of warnings errors.
>> Apart
>>from errors in debugging code an error like this occurs, though in an
>>other file:
>>
>>g++ -c -D_3 -D__PC__ -DDebug -pipe -s -O3 -march=i686
>> -fomit-frame-pointer
>>-funroll-loops -fno-strict-aliasing   -I/usr/X11R6/include
>>-I/home/clenzen/dipl_ware/ug/include -I/home/clenzen/dipl_ware/ug
>>-I../include ugdevices.c
>>In file included from ugdevices.c:40:
>>/home/clenzen/dipl_ware/ug/include/debug.h:176: Fehler: »INT« bezeichnet
>>keinen Typ
>>
>>6. I edited the corresponding "INT" to "int" and had to comment out many
>>debugging lines using an undefinded variable named "me" (marked
>> element?),
>>then ug compiled with standard options. However, installing DUNE the ug
>>libraries were still not accepted (because of a linker error calling some
>>function with grape in its name; all other checks concerning ug were ok).
>>I guessed Grape has to be set to ON in ug.conf and IF = S (concluded from
>>the names of the libraries checked by DUNE's configure). However, this
>>leads to many knew compilation errors for ug. They are like the one
>> above,
>>just with additional other types in capital letters involved. There I
>>stopped trying, because I don't even know what's the purpose of this
>> types
>>(or more correctly, Macros substituting types?) I won't repeat the whole
>>process, if not necessary.
>>
>>7. Compiling your developer version of ug with gcc gives the following
>> error:
>>
>>gcc -c -D_3 -D__PC__ -DDebug -pipe -s -O3 -march=i686
>> -fomit-frame-pointer
>>-funroll-loops -fno-strict-aliasing   -I/usr/X11R6/include
>>-I/home/clenzen/dipl_ware/ug/include -I/home/clenzen/dipl_ware/ug
>>algebra.c
>>In file included from algebra.c:74:
>>refine.h:261: Fehler: Syntaxfehler vor »=«-Zeichen
>>
>>I find this rather confusing, because I don't see the problem with this
>>default value. Editing the INT type to int doesn't change anything (as
>>expected: grep told me that INT is probably just a Makro for int, which
>>againg is confusing - what's the sense in it?). But when "=0" is deleted,
>>it compiles, of course a new error occurs when the function is called
>> with
>>the last argument omitted. Here I stopped again, as ug has to be compiled
>>as c++ code in any case.
>>
>>8. Compiling the developer version of ug with g++ immediatly gives the
>>following error:
>>
>>g++ -c -D_3 -D__PC__ -DDebug -pipe -s -O3 -march=i686
>> -fomit-frame-pointer
>>-funroll-loops -fno-strict-aliasing   -I/usr/X11R6/include
>>-I/home/clenzen/dipl_ware/ug/include -I/home/clenzen/dipl_ware/ug
>>-I../include ugdevices.c
>>In file included from ugdevices.c:40:
>>/home/clenzen/dipl_ware/ug/include/debug.h:176: Fehler: »INT« bezeichnet
>>keinen Typ
>>
>>At this point I didn't try to change anything, because I guess I would
>> get
>>the same problems as above.
>>
>>These errors are the same as on the other system. I hope this is what you
>>requested.
>>
>>Thanks for the help
>>Christoph
>>
>>Oliver Sander wrote:
>>
>>
>>>Dear Christoph!
>>>I'm sorry to hear that you are having difficulties installing UG for
>>>Dune.  We have
>>>had various successful installs by now, but there may still be
>>>difficulties on less
>>>common platforms.
>>>
>>>In order to help you, though, we need some actual information.  What
>>>platform
>>>do you use?  What's your compiler?  And what are the actual error
>>>messages?
>>>Without that kind of information it is hardly possible to pin down any
>>>errors.
>>>
>>>--
>>>Oliver Sander
>>>
>>>Christoph Lenzen wrote:
>>>
>>>
>>>
>>>>Hallo,
>>>>
>>>>ich versuche im Augenblick, die aktuelle Version von UG (ausgecheckt
>>>> mit
>>>>
>>>>cvs -d:pserver:anonymous at sit.iwr.uni-heidelberg.de:/Users/mirror/CVS
>>>>co UG/ug
>>>>) mit der beta5-Version von DUNE zu verheiraten. UG benutzt inzwischen
>>>>eine Datei autogen.sh statt dem vorherigen configure, wie auf der
>>>>DUNE-Seite beschrieben (siehe README.autotools im Rootverzeichnis von
>>>>UG). Ich übergab die angegebenen Optionen, aber DUNE akzeptierte die
>>>>Bibliotheken nicht, da der Linker auf "undefined references" stieß.
>>>>
>>>>Zunächst stellte ich fest, dass UG trotz der übergebenen Option nicht
>>>>g++ statt gcc verwendete (das mag eventuell an meinem System liegen,
>>>>ich wüsste aber nicht wieso). Das habe ich schließlich durch "make
>>>>ARCH_CC=g++" erzwungen, nachdem ich überprüft habe, dass ARCH_CC auf
>>>>gcc gesetzt war. Nach einigen Schwierigkeiten wegen einer INT-Variable
>>>>(mittels grep konnte ich ein #define INT int finden), die ich zu einer
>>>>int gemacht habe, und ausdokumentieren von PRINTDEBUG-Zeilen, die auf
>>>>eine undefinierte Variable me zurückgreifen, wurden die Bibliotheken
>>>>erstellt.
>>>>
>>>>Die Fehlermeldung, die ich bezüglich der Bibliotheken vor der
>>>>Anwendung von g++ in config.log (vom DUNE-configure-Aufruf) fand,
>>>>bezog sich zunächst auf  darauf, dass InitUg(int, int) nicht zu finden
>>>>war. Hinterher erhielt ich eine andere undefinierte Referenz. Ändern
>>>>der Optionen in ug.conf liefert neue Fehler beim Kompilieren, daher
>>>>habe ich Abstand davon genommen, dieses blinde Experimentieren weiter
>>>>zu führen.
>>>>
>>>>Hat jemand eine Idee, was zu tun ist? Sollte ich mich an die
>>>>Programmierer von UG wenden?
>>>>
>>>>Viele Grüße
>>>>Christoph Lenzen
>>>>
>>>>------------------------------------------------------------------------
>>>>
>>>>_______________________________________________
>>>>Dune mailing list
>>>>Dune at dune-project.org
>>>>http://lists.dune-project.org/mailman/listinfo/dune
>>>>
>>>>
>>>>
>>>>
>>>--
>>>************************************************************************
>>>* Oliver Sander                ** email: sander at mi.fu-berlin.de        *
>>>* Freie Universität Berlin     ** phone: + 49 (30) 838 75217           *
>>>* Institut für Mathematik II   ** URL  : page.mi.fu-berlin.de/~sander  *
>>>* Arnimallee 6                 ** -------------------------------------*
>>>* 14195 Berlin, Germany        ** Member of MATHEON (www.matheon.de)   *
>>>************************************************************************
>>>
>>>
>>>
>>>
>>
>>
>>
>
>
> --
> ************************************************************************
> * Oliver Sander                ** email: sander at mi.fu-berlin.de        *
> * Freie Universität Berlin     ** phone: + 49 (30) 838 75217           *
> * Institut für Mathematik II   ** URL  : page.mi.fu-berlin.de/~sander  *
> * Arnimallee 6                 ** -------------------------------------*
> * 14195 Berlin, Germany        ** Member of MATHEON (www.matheon.de)   *
> ************************************************************************
>
>