The R Project SVN R

Rev

Rev 3661 | Blame | Compare with Previous | Last modification | View Log | Download | RSS feed

SUCCESSFUL BUILDS

R is known to configure and compile on the following platforms.
If you have success on any others, let us know the platform names
and any changes you had to make so we can update this file and
our installation procedures.

    PLATFORM            COMPILER    NOTES

    i386-unknown-freebsd2.2.5   gcc / f77
    i386-unknown-linux debian 1.3   gcc / g77
    i386-unknown-linux debian 2.0   egcs/ g77
    i386-unknown-linux Redhat 4.2   egcs/ g77
    i386-unknown-linux Redhat 4.2   gcc / f2c
    i386-unknown-linux RedHat 5.0   (??)
    i386-unknown-linux SuSe 6.0 egcs/ f77
    mips-sgi-irix6.2        cc  / f77   (1)
    sparc-sun-solaris2.5.1      gcc / g77
        "           gcc / f77   (3)
        "           cc  / f77
    sparc-sun-solaris2.6        gcc / f77 
    sparc-sun-solaris2.6        cc  / f77
    alpha-dec-osf4          gcc     (4)

    OLD (pre 0.62 systems)

    hppa1.1-hp-hpux9.07     c89         (2)
    hppa1.1-hp-hpux10.20        c89
    alpha-dec-osf3.2        cc

NOTES

(1) Perl version 5 must be installed in order to build
    the documentation.  (The SGI supplied Perl is version 4).

(2) The Makefile in src/main must be edited manually after
    configuration and the reference to the library "-lSM"
    removed.

(3)     A number of  users appear to have installation problems with f77,
    requiring the f77 library directory to be explicitly 
    specified in -L and -R 

(4)     (Greg Snow <snow@biostat.washington.edu>)

    It needed the -mieee-with-inexact flag (the default config used 
    -ieee_with_inexact, note the m and the switch from _ to -)

    I also changed a line in plot.c, the line was around 411 (I
    added some printf's for debugging so this is probably not the
    original line number)

    the original line was:
    if (FINITE(asp) && asp > 0 ) {
    my new line is:
    if (FINITE(asp) && asp > 0 && asp != NA_REAL ) {

    Apparently my setup thinks that NA_REAL is positive finite so
    the change was neccesary.