[scilab-Users] Building Scilab 5.3.0-beta 4 on Fedora 13
Dean S. Messing
deanm at sharplabs.com
Thu Oct 14 18:47:03 CEST 2010
On Thu, 14 Oct 2010 07:47:41 +0200, Sylvestre Ledru wrote:
> Please, have a look to the output of the config.log which is way more
> precise than the one from the configure (start from the end of this file
> and skip the variable declarations).
> It will tell you what the actual error is.
(Smile)
I've been doing this "in spades" for the last (uncountable number) of
hours. :-)
I've pulled the conftest.c script and gcc lines out of config.log on several
failures to discover what's going on. Sometimes its missing libraries on
my machine. Other times it is what I consider bugs in the script.
Here is the complete list of packages and their dependencies I've had to
pull in since beginning this little project.
From rpm -qa --last:
matio-devel-1.3.4-1.fc13 Thu 14 Oct 2010 09:02:13 AM PDT
matio-1.3.4-1.fc13 Thu 14 Oct 2010 09:02:12 AM PDT
blas-devel-3.2.1-4.fc13 Wed 13 Oct 2010 11:46:37 PM PDT
atlas-devel-3.8.3-16.fc13 Wed 13 Oct 2010 11:26:16 PM PDT
checkstyle-4.1-7.fc12 Wed 13 Oct 2010 10:55:06 AM PDT
xml-commons-apis-javadoc-1.4.01-1.fc13 Wed 13 Oct 2010 10:55:05 AM PDT
ant-javadoc-1.7.1-13.fc13 Wed 13 Oct 2010 10:55:02 AM PDT
antlr-javadoc-2.7.7-6.fc12 Wed 13 Oct 2010 10:55:00 AM PDT
velocity-1.6.3-3.fc13 Wed 13 Oct 2010 10:54:58 AM PDT
jakarta-commons-beanutils-javadoc-1.7.0-12.3.fc12 Wed 13 Oct 2010 10:54:58 AM PDT
jakarta-commons-cli-1.1-5.fc12 Wed 13 Oct 2010 10:54:57 AM PDT
werken-xpath-0.9.4-5.beta.12.3.fc13 Wed 13 Oct 2010 10:54:56 AM PDT
jakarta-commons-collections-3.2.1-3.fc12 Wed 13 Oct 2010 10:54:55 AM PDT
jakarta-commons-beanutils-1.7.0-12.3.fc12 Wed 13 Oct 2010 10:54:55 AM PDT
jdom-1.1.1-1.fc13 Wed 13 Oct 2010 10:54:52 AM PDT
gluegen-1-0.20102502svn11.fc13 Wed 13 Oct 2010 12:30:32 AM PDT
libCg-2.2.0017-1.fc13 Tue 12 Oct 2010 11:45:29 PM PDT
saxon-9.2.0.3-1.fc13 Tue 12 Oct 2010 11:41:11 PM PDT
bea-stax-1.2.0-0.5.rc1.fc12 Tue 12 Oct 2010 11:41:10 PM PDT
bea-stax-api-1.2.0-0.5.rc1.fc12 Tue 12 Oct 2010 11:41:08 PM PDT
pcre-devel-7.8-3.fc12 Tue 12 Oct 2010 11:40:19 PM PDT
suitesparse-devel-3.4.0-2.fc12 Tue 12 Oct 2010 11:31:54 PM PDT
fftw-devel-3.2.2-4.fc13 Tue 12 Oct 2010 11:31:54 PM PDT
pvm-3.4.5-14.fc12 Tue 12 Oct 2010 11:31:53 PM PDT
tcsh-6.17-6.fc13 Tue 12 Oct 2010 11:31:51 PM PDT
ocaml-3.11.2-2.fc13 Tue 12 Oct 2010 11:26:55 PM PDT
ocaml-runtime-3.11.2-2.fc13 Tue 12 Oct 2010 11:26:53 PM PDT
tk-devel-8.5.8-2.fc13 Tue 12 Oct 2010 11:26:33 PM PDT
tcl-devel-8.5.8-1.fc13 Tue 12 Oct 2010 11:26:32 PM PDT
jeuclid-3.1.3-12.fc12 Tue 12 Oct 2010 11:25:05 PM PDT
jcip-annotations-1-1.20060626.fc13 Tue 12 Oct 2010 11:25:05 PM PDT
javahelp2-2.0.05-8.fc12 Tue 12 Oct 2010 11:22:48 PM PDT
jrosetta-1.0.2-3.fc12 Tue 12 Oct 2010 11:12:33 PM PDT
cpptasks-1.0b5-5.fc12 Tue 12 Oct 2010 10:43:22 PM PDT
Ok, these were my problem.
But other problems are with the Scilab ./configure script itself.
For example, to fix my first "jogl" problem (see the 1st e-mail of this
thread), I moved the contents of "bin/" from
prerequirements-scilab-5.3.0-beta-4 into /usr/local/lib64.
(I had already moved "thirdparty/" from there into the Build directory,
but since it already had a bin directory, I got confused.
That solution fixed the missing "jogl" depdendency. It was obviously
finding "libjogl.so" and its cousins. But then ./configure died from
not finding "libjhdf5.so", which you helped me diagnose from the cryptic
messages. Problem is "libjhdf5.so" was right there in /usr/local/lib64
along with "libjogl.so"!
I consider this a "library referencing" bug in the script, of which I've
discovered several more (see below). Anyway, I finally discovered that
by moving the contents of "bin/" into the "bin" directory of the Build
dir, both problems were solved.
Next came the PVM problems. Again, the cryptic messages totally confused
me at first. As it turns out, ./configure was unable to execute
"pvmgetarch" in /usr/share/pmv3/lib, even though it found the directory
just fine. I ended up having to put "/usr/share/pmv3/lib" in my $PATH
to get past this. That seems like a bug too.
The rest (so far--I've still not made it all the way
through ./configure !) of my problems have had to do with missing -L
libraries on the various "conftest.c" compiles and links.
I discovered these by pulling out conftest.c and the compile command
and runing it on the commandline, then adding the missing libraries.
I'm now having to use
./configure --with-blas-library=/usr/lib64/atlas\
--with-lapack-library=/usr/lib64/atlas\
--with-umfpack-library=/usr/lib64/atlas
to get through these. Each flag was needed to fix a different failure.
If I leave them all off this is the output of ./configure when it dies:
checking if BLAS, ATLAS or MKL is available...
checking how to get verbose linking output from gfortran... -v
checking for Fortran 77 libraries of gfortran... -L/usr/lib/gcc/x86_64-redhat-linux/4.4.4 -L/usr/lib/gcc/x86_64-redhat-linux/4.4.4/../../../../lib64 -L/lib/../lib64 -L/usr/lib/../lib64 -L/usr/lib/gcc/x86_64-redhat-linux/4.4.4/../../.. -lpthread -ldl -lcurses -lgfortranbegin -lgfortran -lm
checking for dummy main to link with Fortran 77 libraries... none
checking for Fortran 77 name-mangling scheme... lower case, underscore, no extra underscore
checking for sgemm_... no
checking for ATL_xerbla in -latlas... no
checking for sgemm_ in -lmkl... no
checking for sgemm_ in -lblas... yes
checking for dgemm_ in -ldgemm... no
checking for sgemm_ in -lcxml... no
checking for sgemm_ in -ldxml... no
checking for sgemm_ in -lscs... no
checking for sgemm_ in -lcomplib.sgimath... no
checking for sgemm_ in -lblas... (cached) yes
checking for sgemm_ in -lessl... no
checking for sgemm_ in -lblas... (cached) yes
Generic Blas found
checking if LAPACK is available...
checking for cheev_... no
checking for cheev_ in -llapack... no
checking for cheev_ in -llapack_rs6k... no
configure: error: Impossible to find the LAPACK library.
scilab-master-1286272063=>
Yet, both lapack and atlas are installed on my system.
The atlas libs are in /usr/lib64/atlas and the lapack libs
are in /usr/lib64.
Seems like the configure script should be able to find these.
That's where I am now. I hope to get through this soon and actually
build scilab and use it to do some productive work. :-)
Regrds,
Dean
More information about the users
mailing list