[Rd] R 1.4.0 build fails on AIX (PR#1289)

Thomas Lumley tlumley@u.washington.edu
Wed, 30 Jan 2002 08:45:53 -0800 (PST)

On Wed, 30 Jan 2002 lio@hpss1.ccs.ornl.gov wrote:

> Full_Name: Daniel L. Million
> Version: 1.4.0
> OS: AIX 5.1 and 4.3.3
> Submission from: (NULL) (
> Is there any guidance available on building R on AIX?  Specifically, which C
> compiler to use?  We have AIX XL Fortran version 7.1.1.  We also have AIX
> Visual Age C version 5.0.2.  We also have gcc available.
> I have tried building with "cc" (the default AIX compiler) and "gcc".  During
> the "make", C compilation in src/library/methods/src completes, and then I
> see this:
> mkdir ../../../../library/methods/libs
> dumping R code in package `methods'
> make: 1254-059 The signal code from the last command is 11.
> Apparently the make is running R.bin, which coredumps with a segmentation
> violation.
> Do I need to be using a thread-safe compiler (e.g., "cc_r") for this build?
> Any clues at all as to how to get around this?

I have successfully built most recent versions of R including 1.4.0 with
gcc/g77 and with "C for AIX version 5"/"XL Fortran for AIX" on AIX 4.3. I
haven't been able to mix the GNU and IBM compilers, but I haven't tried
recently.  Last week I compiled R-patched successfully using gcc/g77. I
didn't have to do anything special, but AIX has often been source of
configuration bugs.

You could try not building the methods package: edit  Makeconf in the top
R/ directory and remove methods from R_PKGS.

Also, I think that it's possible to have the methods package build in the
usual source form rather than as a saved image. I don't know how, but John
Chambers or Duncan Temple Lang may be able to help.

These wouldn't actually fix the problem, but would enable you to compile
everything else.

I'll try again with the AIX compilers and the new R 1.4.1. It's possible
that there have been some changes since I last tried, since I usually do
this shortly before a release.


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