[[0.62.3, already 0.62.2]] This bug report is overdue, but I really didn't test these things for weeks (have always used gcc, but only yesterday, someone told me that he saw a speed gain of a factor 2 when using Sun's cc over gcc) I just found that the same problem is already in 0.62.2 If I take yesterday's R-release.tar.gz (or also R-0.62.2.tar.gz) unpack and add CC=cc ----- to config.site, then ./configure ----------- it ends VERY badly: % ./configure creating cache ./config.cache checking for a BSD compatible install... /usr/local/bin/ginstall -c checking whether ln -s works... yes checking for ranlib... ranlib checking for bison... bison -y checking for ar... ar checking for ratfor... ratfor checking for latex... /afs/ethz.ch/public/teTeX/teTeX/bin/sparc-solaris2.5/latex checking for dvips... /afs/ethz.ch/public/teTeX/teTeX/bin/sparc-solaris2.5/dvips checking for perl... /usr/local/bin/perl checking whether perl is perl 5... yes checking for cc... cc checking for ranlib... (cached) ranlib checking for f77... f77 checking for underscore after Fortran symbols... configure: error: Nothing worked - cannot use FORTRAN ----------- and then STOP, nothing, .... When I omit the "CC=cc", gcc is used (together with f77), and all is well. % cc -V cc: SC3.0.1 13 Jul 1994 usage: cc [ options] files. Use 'cc -flags' for details % f77 -V f77: SC3.0.1 13 Jul 1994 Usage: f77 [ options ] files. Use 'f77 -flags' for details % uname -a SunOS florence 5.5 Generic sun4u sparc SUNW,Ultra-1 ------------------------------------------------ So yes, the SC.. compilers are somewhat dated, but still... -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- 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 _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Martin Maechler
1998-Aug-18 13:54 UTC
Problem in "configure" for Solaris (cc) -- solved (partly) --
(answering myself, after a suggestion by Kurt Hornik which actually was *not* the solution.. but...) The configure problem I've just reported (( setting CC=cc in config.site )) actually does not show when I either leave the CFLAGS commented out (which results in CFLAGS=-g) or set CFLAGS="-g -O", however, it *DOES* show when I set CFLAGS=-O2 or CFLAGS="-g -O2" ---- Reading the long output of ``cc -flags'', I see that more than average optimization is done using cc -xO[1-4] # Still very funny that "-O2" leads to such bad results with ./configure. Maybe configure could become smarter here... Martin -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- 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 _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._