R devel - Jul 2002 - compilation breaks with gnome support on hp-ux 11.00 (PR#1757)

Full_Name: Martin Gansser
Version: 1.5.1
OS: hp-ux 11.00
Submission from: (NULL) (195.145.22.147)


cc -Wp,-H16000 -I. -I../../../src/include -I../../../src/include
-I/opt/gnome/include/gnome-1.0 -DNE
ED_GNOMESUPPORT_H -I/opt/gnome/lib/gnome-libs/include
-I/opt/gnome/include/glib-1.2 -I/opt/gnome/lib
/glib/include -I/opt/gnome/include/orbit-1.0 -I/opt/gnome/include/gtk-1.2
-I/opt/gnome/include/gnome
-xml -I/opt/gnome/include -I/opt/gnome/include/gnome-1.0 -DNEED_GNOMESUPPORT_H
-I/opt/gnome/lib/gnom
e-libs/include -I/opt/gnome/include/glib-1.2 -I/opt/gnome/lib/glib/include
-I/opt/gnome/include/orbi
t-1.0 -I/opt/gnome/include/gtk-1.2  -I/usr/local/include -I/opt/gnome/include
-DHAVE_CONFIG_H  +Z  -
Wp,-H16000 -c system-showfiles.c -o system-showfiles.lo
cpp: "/opt/gnome/include/gnome-1.0/libgnome/gnome-i18n.h", line 10:
warning
2013: Unknown preprocess
ing directive.
cc: "system-showfiles.c", line 271: error 1502: Array size must be a
constant
expression.
gmake[4]: *** [system-showfiles.lo] Error 1

gnome-1.4 for hpux 11.00
PS: R-1.5.1 compiles fine w/o gnome support

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

mgansser@rand.de writes:
> cc: "system-showfiles.c", line 271: error 1502: Array size must
be a constant
> expression.
> gmake[4]: *** [system-showfiles.lo] Error 1
> 
> gnome-1.4 for hpux 11.00
> PS: R-1.5.1 compiles fine w/o gnome support
Hmm:

  const gint bufsize = 2048;
  gchar buf[bufsize];

is the culprit, so a quick fix should be obvious (just use #define
instead). But is this usage of "const" variables an un-ANSI C
extension in GCC, or is it the HP C compiler that is to blame??

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard@biostat.ku.dk)             FAX: (+45) 35327907
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

> mgansser@rand.de writes:
> 
> > cc: "system-showfiles.c", line 271: error 1502: Array size 
> must be a constant
> > expression.
> > gmake[4]: *** [system-showfiles.lo] Error 1
> > 
> > gnome-1.4 for hpux 11.00
> > PS: R-1.5.1 compiles fine w/o gnome support
> 
> Hmm:
> 
>   const gint bufsize = 2048;
>   gchar buf[bufsize];
> 
> is the culprit, so a quick fix should be obvious (just use #define
> instead). But is this usage of "const" variables an un-ANSI C
> extension in GCC, or is it the HP C compiler that is to blame??
> 
> -- 
>    O__  ---- Peter Dalgaard             Blegdamsvej 3  
>   c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
>  (*) \(*) -- University of Copenhagen   Denmark      Ph: 
> (+45) 35327918
> ~~~~~~~~~~ - (p.dalgaard@biostat.ku.dk)             FAX: 
> (+45) 35327907
occurs only, if CC=cc (HP C compiler) is set, not with CC=gcc
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

> Hmm:
>
> const gint bufsize = 2048;
> gchar buf[bufsize];
>
> is the culprit, so a quick fix should be obvious (just use #define
> instead). But is this usage of "const" variables an un-ANSI C
> extension in GCC, or is it the HP C compiler that is to blame?? 
Peter (and others):

AFAIK, gcc is the only C compiler that allows this sort of fortran-like syntax.
Mighty handy, but not portable at all. However, behind the screens, it
translates this (more or less) into a call to alloca(). This is a function that
is part of most modern systems, and of which portable versions can be found
(outside the glibc source tree). BTW: debugging code that allocates arrays like
this is not trivial, as on all systems I've seen, you cannot directly access
element n as array[n] (from within gdb, that is).

There is a comprehensive whitepaper about gcc's not-portable-outside-gcc
quirks available on Metrowerk's site, called Porting GNU C Programs to
Metrowerks' CodeWarrior C Compiler.

RenE
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._