On 15 September 2011 at 10:47, Stefano Bridi wrote:
| Sorry... I sent the previous email by error while I was writing ;)
|
| Without reingeneering all the build process I also think a check on
| uname -m could be sufficient.
|
| if [ $(uname -m) == "x86_64" ]; then
| CPPFLAGS="-DWL=64 -DFORTRANUNDERSCORE $CPPFLAGS";
| else
| CPPFLAGS="-DWL=32 -DFORTRANUNDERSCORE $CPPFLAGS";
| fi
Yes. I wonder, though, if we should go one further and create a program like
#include <stdio.h>
int main(void) {
printf("%u\n", (unsigned) sizeof(long));
}
which prints 4 and 8, respectively, on my 32 and 64 bit systems. I may mod it
to print
printf("-DWL=%u\n", 8*(unsigned) sizeof(long));
and that would generate the appropriate switch, also catching other 64 bit
systems and not just the Intel CPU based ones.
Or maybe test for sizeof==8 and then print 64, else print 32 as a fallback.
Dirk
|
| Thanks
| Stefano
|
|
| On Thu, Sep 15, 2011 at 10:45 AM, Stefano Bridi <[email protected]>
wrote:
| > Without reingeneering all the build process I also think a check on
| > uname -m could be sufficient.
| >
| > if [ $(uname -m) == "x86_64" ]; then
| > CPPFLAGS="-DWL=32 -DFORTRANUNDERSCORE $CPPFLAGS";
| > else
| > cho "64bit"; else echo "other";fi
| >
| >
| > On Wed, Sep 14, 2011 at 2:02 PM, Dirk Eddelbuettel <[email protected]> wrote:
| >>
| >> On 14 September 2011 at 12:33, Stefano Bridi wrote:
| >> | Package: pgapack
| >> | Version: 1.1.1-2
| >> | Severity: important
| >> |
| >> | Dear Maintainer,
| >> |
| >> | I was trying to compile the pgapack from source and I always got the
embedded
| >> | test to fail. After some try and error i found that the parameter
"-DWL=32" in
| >> | the configure/configure.in cause the first and last test to fail on my
64bit pc
| >> | ("-DWL=64" seems to be fine).
| >> | So i tryied the packaged library and found that the problem is also here
and so
| >> | this bug report:
| >> | In fact: using the test (instverf) in the pgapack package build against
the
| >> | "libpgapack-serial1" package library I see the same behaviour:
| >> | Test 0 had 994 errors.
| >> | Test 1 was successful.
| >> | Test 2 was successful.
| >> | Test 3 was successful.
| >> | Test 4 had 989 errors.
| >> | With a rebuild from source of the same library with the "-DWL=64" I get
| >> | Test 0 was successful.
| >> | Test 1 was successful.
| >> | Test 2 was successful.
| >> | Test 3 was successful.
| >> | Test 4 was successful.
| >> |
| >> | I'm the system administrator and not the end user of this library and so
I
| >> | don't know what is going on ...
| >>
| >> The sources are about 15 years old, and were unmaintained for many years
| >> until I adopted them.
| >>
| >> I have not changed the build process. For Linux, it does an unconditional
| >>
| >> linux)
| >> SHELL="/bin/sh"
| >> CC=cc
| >> FC=f77
| >> FFLAGS="-w"
| >> LDFLAGS="-s $LDFLAGS"
| >> CPPFLAGS="-DWL=32 -DFORTRANUNDERSCORE $CPPFLAGS"
| >> ;;
| >>
| >> We could add a simple word-length test here -- this is shell script code.
Do
| >> you have a suggestion, besides maybe testing 'uname -m' for 'x86_64' and
then
| >> setting -DWL=64 ?
| >>
| >> I don't really want to re-engineer the whole build system....
| >>
| >> Dirk
| >>
| >>
| >> | -- System Information:
| >> | Debian Release: wheezy/sid
| >> | APT prefers unstable
| >> | APT policy: (500, 'unstable'), (1, 'experimental')
| >> | Architecture: amd64 (x86_64)
| >> |
| >> | Kernel: Linux 3.0-3.dmz.2-liquorix-amd64 (SMP w/2 CPU cores; PREEMPT)
| >> | Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
| >> | Shell: /bin/sh linked to /bin/dash
| >> |
| >> | pgapack depends on no packages.
| >> |
| >> | pgapack recommends no packages.
| >> |
| >> | Versions of packages pgapack suggests:
| >> | ii libpgapack-mpi1 1.1.1-2
| >> | ii libpgapack-serial1 1.1.1-2
| >> |
| >> | -- no debconf information
| >> |
| >> |
| >>
| >> --
| >> New Rcpp master class for R and C++ integration is scheduled for
| >> San Francisco (Oct 8), more details / reg.info available at
| >>
http://www.revolutionanalytics.com/products/training/public/rcpp-master-class.php
| >>
| >
--
New Rcpp master class for R and C++ integration is scheduled for
San Francisco (Oct 8), more details / reg.info available at
http://www.revolutionanalytics.com/products/training/public/rcpp-master-class.php
--
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]
