[R] "make check" errors when checking R1.9.1 build (was RE: Building V2.3.0 on Tru64 V5.1B)

[Ravi, Narendra, INFOT]

 Hi,
	I am attempting to build R on Alpha servers running Tru64 UNIX.
I am having problems building R 2.3.0 so am attempting to build older
releases.

	I have been able to build R 1.9.1 without problem. However "make
check" fails for the following tests:

1. arith.Rout
The difference in the output is as follows:

# diff arith.Rout.save ../../tests/arith.Rout.save
43,46c43,46
<   -4 0.000-0.500i 1+0i 0.000+2.000i -4+0i -0.000-8.000i  16+-0i
<   -3 0.000-0.577i 1+0i 0.000+1.732i -3+0i -0.000-5.196i   9+-0i
<   -2 0.000-0.707i 1+0i 0.000+1.414i -2+0i -0.000-2.828i   4+-0i
<   -1 0.000-1.000i 1+0i 0.000+1.000i -1+0i -0.000-1.000i   1+-0i
---
>   -4 0.000-0.500i 1+0i 0.000+2.000i -4+0i  0.000-8.000i  16+0i
>   -3 0.000-0.577i 1+0i 0.000+1.732i -3+0i  0.000-5.196i   9+0i
>   -2 0.000-0.707i 1+0i 0.000+1.414i -2+0i  0.000-2.828i   4+0i
>   -1 0.000-1.000i 1+0i 0.000+1.000i -1+0i  0.000-1.000i   1+0i
66c66
< [1] -0.0720852+0.6383267i -0.6058296+-0.0000000i -0.0720852-0.6383267i
---
> [1] -0.0720852+0.6383267i -0.6058296+0.0000000i -0.0720852-0.6383267i
135c135
< n= 30 : 465+0i -15+142.7155i -15+70.56945i -15+46.16525i -15+33.69055i
-15+25.98076i -15+20.64573i -15+16.65919i -15+13.50606i -15+10.89814i
-15+8.660254i -15+6.67843i -15+4.873795i -15+3.188348i -15+1.576564i
-15+-0i -15-1.576564i -15-3.188348i -15-4.873795i -15-6.67843i
-15-8.660254i -15-10.89814i -15-13.50606i -15-16.65919i -15-20.64573i
-15-25.98076i -15-33.69055i -15-46.16525i -15-70.56945i -15-142.7155i
---
> n= 30 : 465+0i -15+142.7155i -15+70.56945i -15+46.16525i -15+33.69055i
-15+25.98076i -15+20.64573i -15+16.65919i -15+13.50606i -15+10.89814i
-15+8.660254i -15+6.67843i -15+4.873795i -15+3.188348i -15+1.576564i
-15+0i -15-1.576564i -15-3.188348i -15-4.873795i -15-6.67843i
-15-8.660254i -15-10.89814i -15-13.50606i -15-16.65919i -15-20.64573i
-15-25.98076i -15-33.69055i -15-46.16525i -15-70.56945i -15-142.7155i

2. d-p-q-r-tests.Rout

# diff d-p-q-r-tests.Rout.save ../../tests/d-p-q-r-tests.Rout.save
405c405
< [20,] 40  0.000000e+00   -0.000000e+00           -Inf      -804.608442
---
> [20,] 40  0.000000e+00    0.000000e+00           -Inf      -804.608442


3. Regression Tests:

running regression tests
make[3]: Entering directory
`/usr/ruby-deploy/swkits/RUBY/SOFTWARE/3rdPartySW/R/R-1.9.1/build/tests'
running code in '../../tests/reg-tests-1.R'
...228814:/usr/ruby-deploy/swkits/RUBY/SOFTWARE/3rdPartySW/R/R-1.9.1/bui
ld/bin/R.bin: /sbin/loader: Fatal Error: call to unresolved symbol from
/usr/ruby-deploy/swkits/RUBY/SOFTWARE/3rdPartySW/R/R-1.9.1/build/modules
/lapack.so (pc=0x3ffbfde3528)
make[3]: *** [reg-tests-1.Rout] Error 1


I am a neophyte when it comes to using R and would appreciate any
insight into what may be the problem.

Narendra Ravi

-----Original Message-----
From: Ravi, Narendra, INFOT 
Sent: Friday, May 26, 2006 7:50 PM
To: 'r-help at stat.math.ethz.ch'
Subject: Building V2.3.0 on Tru64 V5.1B

Hi,
	I am going to be attempting a build of R for the alphas (our
main need is to collate and present graphically, performance metrics we
gather on our applications).

We have:

CC=cc, CXX=cxx, Fortran V5.51 (f77, f95), GNU make 3.80, Perl 5.8.4

I am hesitant to rebuild GCC on this platform, and would like to know if
others have success in building R with native C, C++ and Fortran. I plan
on using the following configure string:

CPPFLAGS="-I/opt/gnu/include -I/opt/libjpeg/include
-I/opt/libpng/include -I/opt/ncurses/include" CFLAGS="-std1"
LDFLAGS="-L/opt/gnu/lib -L/opt/libjpeg/lib -L/opt/libpng/lib
-L/opt/ncurses/lib -lncurses -ljpeg -lpng -rpath
"/opt/libpng/lib:/opt/libjpeg/lib:/opt/ncurses/lib:/opt/gnu/lib:/usr/shl
ib" R_PAPERSIZE=letter CC=cc CXX=cxx ./configure prefix=/opt/R
--without-readline --without-blas

I expect to update the config.site file with the above values. If
someone has built R-2.3.0 on Tru64 V5.1B, I would appreciate hearing
how.

Thank you,

Narendra Ravi