search

MichaelChirico / r-bugs

A ⚠️read-only⚠️mirror of https://bugs.r-project.org/
20 stars 0 forks source link

[BUGZILLA #7351] Comments on the R-2.0.0 release #2287

Closed MichaelChirico closed 4 years ago

MichaelChirico commented 4 years ago

From: "Nelson H. F. Beebe" <beebe@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>> Builds of the R-2.0.0 release have been considerably more successful at my site than previous releases. I now have it installed on these platforms:

Apple PowerPC G3 267MHz GNU/Linux 2.4.19-4a (Yellow Dog Linux release 2.3 (Dayton))
Compaq Alpha Sierra OSF/1 5.1
Compaq/DEC Alpha    OSF/1 4.0F
Intel Itanium-2     GNU/Linux Red Hat Linux Advanced Server release 2.1AS (Derry)
Intel Pentium II    FreeBSD 4.4-RELEASE #0
Intel Pentium III   FreeBSD 5.0
Intel Pentium III   GNU/Linux 2.4.18-26.8.0smp (Red Hat 8.0 (Psyche))
Intel Pentium III   GNU/Linux 2.4.21-20.ELsmp (Red Hat Enterprise Linux WS release 3 (Taroon Update 3))
Intel Pentium III   NetBSD 1.6
Intel Pentium III   OpenBSD 3.2
Intel Pentium III   Solaris 9 x86
SGI Origin 200      IRIX 6.5
Sun SPARC       GNU/Linux 2.2.19-6.2.1 (Red Hat 6.2)
Sun SPARC       Solaris 2.7
Sun SPARC       Solaris 2.8
Sun SPARC       Solaris 2.9

However, builds initially failed on several of these systems because of errors like this:

/usr/bin/ld: cannot find -lX11

On these systems, I was able to restart the build with

make LDFLAGS=-L/usr/X11R6/lib

and have it complete successfully. I've recently seen some other packages that failed to compile for the same reason: the list of standard locations that is searched at configure time for X11 libraries is out-of-date. The R-2.0.0 configure.ac script has this snippet:

for dir in /usr/bin/X11 /usr/X11R6/bin /usr/openwin/bin; do
  r_xtra_path="${r_xtra_path}${PATH_SEPARATOR}${dir}"
done

However, it does not seem to have an analogue for the corresponding lib directory.

Also, on some systems, I had to switch from using the native C/C++ compilers to GNU gcc/g++.

R-2.0.0 has failed to build on these systems:

Apple PowerPC G4 1.4GHz Darwin 7.3.0 (MacOS 10.3.2 (7D26))
SGI Indy-200        GNU/Linux Gentoo 1.4.16
IBM PowerPC     AIX 4.2

On the MacOS 10 system, builds with the native cc (== gcc 3.3 20030304) and with the latest gcc-3.4.3 fail like this:

gcc -dynamiclib -L/usr/local/lib -dynamiclib -install_name libR.dylib \
    -compatibility_version 2.0.0 -current_version 2.0.0 \
    -headerpad_max_install_names -framework Carbon -o libR.dylib \
    Rembedded.lo CConverters.lo CommandLineArgs.lo Rdynload.lo Renviron.lo \
    RNG.lo apply.lo arithmetic.lo apse.lo array.lo attrib.lo base.lo \
    bind.lo builtin.lo character.lo coerce.lo colors.lo complex.lo \
    connections.lo context.lo cov.lo cum.lo dcf.lo datetime.lo debug.lo \
    devPS.lo devPicTeX.lo deparse.lo deriv.lo devices.lo dotcode.lo \
    dounzip.lo dstruct.lo duplicate.lo engine.lo envir.lo errors.lo \
    eval.lo format.lo fourier.lo gram.lo gram-ex.lo graphics.lo \
    identical.lo internet.lo iosupport.lo lapack.lo list.lo logic.lo \
    main.lo mapply.lo match.lo memory.lo model.lo names.lo objects.lo \
    optim.lo optimize.lo options.lo par.lo paste.lo pcre.lo platform.lo \
    plot.lo plot3d.lo plotmath.lo print.lo printarray.lo printvector.lo \
    printutils.lo qsort.lo random.lo regex.lo registration.lo relop.lo \
    saveload.lo scan.lo seq.lo serialize.lo size.lo sort.lo source.lo \
    split.lo sprintf.lo startup.lo subassign.lo subscript.lo subset.lo \
    summary.lo sysutils.lo unique.lo util.lo version.lo vfonts.lo xxxpr.lo \
    `ls ../appl/*.lo ../nmath/*.lo ../unix/*.lo 2>/dev/null` -lblas \
    -L/usr/local/lib \
    -L/export/local/powerpc-macosx/bin/../lib/gcc/powerpc-apple-darwin7.5.0/3.4.2 \
    -L/export/local/powerpc-macosx/bin/../lib/gcc \
    -L/usr/local/lib/gcc/powerpc-apple-darwin7.5.0/3.4.2 \
    -L/export/local/powerpc-macosx/bin/../lib/gcc/powerpc-apple-darwin7.5.0/3.4.2/../../..  \
    -L/usr/local/lib/gcc/powerpc-apple-darwin7.5.0/3.4.2/../../.. -lfrtbegin \
    -lg2c -lgcc_s -lSystem ../extra/zlib/libz.a ../extra/bzip2/libbz2.a \
    ../extra/pcre/libpcre.a -lreadline -lcc_dynamic -lncurses -lm \
    -lcc_dynamic \
gcc: Carbon: No such file or directory

On the GNU/Linux MIPS system, there is as yet no working Fortran compiler, and f2c fails to build there, so R cannot be expected to run.

On the IBM AIX system, I get this failure:

gcc -Wl,-bM:SRE -Wl,-H512 -Wl,-T512 -Wl,-bnoentry -Wl,-bexpall \
    -Wl,-bI:../../../etc/R.exp -Wl,-L/usr/local/lib \
    -Wl,-bE:../../../etc/Rlapack.exp -o libRlapack.so dlamc.lo \
    dlapack0.lo dlapack1.lo dlapack2.lo dlapack3.lo cmplx.lo \
    cmplxblas.lo \
    -L/usr/local/lib/gcc-lib/powerpc-ibm-aix4.2.1.0/2.95.3 \
    -L/usr/local/powerpc-ibm-aix4.2.1.0/lib -L/usr/local/lib -lg2c \
    -lm \
    /usr/local/lib/gcc-lib/powerpc-ibm-aix4.2.1.0/2.95.3/libgcc.a \
    -lg
ld: 0711-224 WARNING: Duplicate symbol: .lsame_
ld: 0711-224 WARNING: Duplicate symbol: lsame_
ld: 0711-345 Use the -bloadmap or -bnoquiet option to obtain more information.
make[4]: Leaving directory `/local/build/R-2.0.0/src/modules/lapack'

...

gcc -Wl,-bM:SRE -Wl,-H512 -Wl,-T512 -Wl,-bnoentry -Wl,-bexpall \
    -Wl,-bI:../../../etc/R.exp -Wl,-L/usr/local/lib -o lapack.so \
    -Wl,-bI:../../../etc/Rlapack.exp Lapack.lo rgeev.lo rsyev.lo \
    -L../../../lib -lRlapack \
    -L/usr/local/lib/gcc-lib/powerpc-ibm-aix4.2.1.0/2.95.3 \
    -L/usr/local/powerpc-ibm-aix4.2.1.0/lib -L/usr/local/lib -lg2c -lm \
    /usr/local/lib/gcc-lib/powerpc-ibm-aix4.2.1.0/2.95.3/libgcc.a -lg \
    -lreadline -ldl -lncurses -lm -lc \
    ld: 0706-006 Cannot find or open library file: -l Rlapack \
    ld:open(): No such file or directory
    collect2: ld returned 255 exit status
    make[4]: Leaving directory `/local/build/R-2.0.0/src/modules/lapack'

Shared libraries (*.so) are uncommon on AIX 4.2; there are none at all in /usr/lib or /lib or /usr/lpp/X11/lib/, and only a few locally-added ones in /usr/local/lib.

------------------------------------------------------------------------------- <CENSORING FROM DETECTED PHONE NUMBER ONWARDS; SEE BUGZILLA>

METADATA

MichaelChirico commented 4 years ago

From: Dirk Eddelbuettel <edd@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>> Nelson,

That is an impressive list of successes.

On Mon, Nov 08, 2004 at 07:41:11PM +0100, beebe@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::> wrote: [...]

SGI Indy-200            GNU/Linux Gentoo 1.4.16

[...]

On the GNU/Linux MIPS system, there is as yet no working Fortran
compiler, and f2c fails to build there, so R cannot be expected to
run.

We have that working under Debian using our standard g77 builds. Logs for mips (and mipsel) since R 1.4.0:

http://buildd.debian.org/build.php?pkg=r-base&arch=mips http://buildd.debian.org/build.php?pkg=r-base&arch=mipsel

I could of course be off-base here if that particular Indy model uses a different cpu -- apologies if that is the case.

Without the trailing &arch=mips(el), it will list logs for all arches.

Dirk

-- If your hair is standing up, then you are in extreme danger. -- http://www.usafa.af.mil/dfp/cockpit-phys/fp1ex3.htm

METADATA

MichaelChirico commented 4 years ago

NOTES: Not sure there is a bug report anywhere in here. Maybe a suggestion to add /usr/X11R6/lib to the LDFLAGS?

METADATA

MichaelChirico commented 4 years ago

Audit (from Jitterbug): Tue Nov 9 08:56:35 2004 ripley changed notes Tue Nov 9 08:56:35 2004 ripley moved from incoming to not-reproducible

METADATA

MichaelChirico commented 4 years ago

From: Prof Brian Ripley <ripley@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>> I don't see the bug report in here. What is the bug you wished to report? (That R requires some flags set on some platforms is documented in the R-admin manual, and not a bug.)

Comments are better sent to the mailing list, not the bug repository, as someone now has to decide what to do with this.

On Mon, 8 Nov 2004 beebe@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::> wrote:

Builds of the R-2.0.0 release have been considerably more successful
at my site than previous releases.  I now have it installed on these
platforms:

Apple PowerPC G3 267MHz GNU/Linux 2.4.19-4a (Yellow Dog Linux release
2.3 (Dayton))
Compaq Alpha Sierra     OSF/1 5.1
Compaq/DEC Alpha        OSF/1 4.0F
Intel Itanium-2         GNU/Linux Red Hat Linux Advanced Server release
2.1AS (Derry)
Intel Pentium II        FreeBSD 4.4-RELEASE #0
Intel Pentium III       FreeBSD 5.0
Intel Pentium III       GNU/Linux 2.4.18-26.8.0smp (Red Hat 8.0
(Psyche))
Intel Pentium III       GNU/Linux 2.4.21-20.ELsmp (Red Hat Enterprise
Linux WS release 3 (Taroon Update 3))
Intel Pentium III       NetBSD 1.6
Intel Pentium III       OpenBSD 3.2
Intel Pentium III       Solaris 9 x86
SGI Origin 200          IRIX 6.5
Sun SPARC               GNU/Linux 2.2.19-6.2.1 (Red Hat 6.2)
Sun SPARC               Solaris 2.7
Sun SPARC               Solaris 2.8
Sun SPARC               Solaris 2.9

However, builds initially failed on several of these systems because
of errors like this:

/usr/bin/ld: cannot find -lX11

On these systems, I was able to restart the build with

make LDFLAGS=-L/usr/X11R6/lib

and have it complete successfully.  I've recently seen some other
packages that failed to compile for the same reason: the list of
standard locations that is searched at configure time for X11
libraries is out-of-date.  The R-2.0.0 configure.ac script has this
snippet:

for dir in /usr/bin/X11 /usr/X11R6/bin /usr/openwin/bin; do
r_xtra_path="${r_xtra_path}${PATH_SEPARATOR}${dir}"
done

However, it does not seem to have an analogue for the corresponding
lib directory.

That's because configure --help says

X features: --x-includes=DIR X include files are in DIR --x-libraries=DIR X library files are in DIR

so I think you omitted --x-libraries=/usr/X11R6/lib. This is also documented in the R-admin manual.

Note that including /usr/X11R6/lib would be rather dangerous, as on 64-bit Linux systems the correct directory is (at least on ours)
/usr/X11R6/lib64. I believe all properly configured systems will have the appropriate one of /usr/X11R6/lib64, /usr/X11R6/lib in the default library path, but if not it will not be a good idea for R to start guessing.

[...]

-- Brian D. Ripley, ripley@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::> Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ <CENSORING FROM DETECTED PHONE NUMBER ONWARDS; SEE BUGZILLA>

METADATA