Well, I am at a loss... Just after I sent this message to you, I tried one more time on the Powerbook, and now it appears to be successfully executing getBioC. I wish I new what I did differently. So far, it appears to be compiling successfully. I won't know until a lot later. I don't think the bioconductor default installations use g77. I need to install DNAcopy which is where I had problems in the past. I will have to wait to try this again on the XServe. It is down at the moment for some hardware work. Thanks Mike --- -------- Original Message -------- Subject: Re: [R-SIG-Mac] Installing Bioconductor 1.5 on R 2.0 for OS X Date: Mon, 08 Nov 2004 16:21:19 -0600 From: Michael Redmond <redmond@cs.wisc.edu> Reply-To: redmond@cs.wisc.edu Organization: Dept. of Statistics, UW-Madison To: Byron Ellis <ellis@stat.harvard.edu> CC: r-sig-mac@stat.math.ethz.ch References: <418805F8.3080505@stat.wisc.edu> <A9350DA4-2D1F-11D9-9AC4-000D93C24D64@stat.harvard.edu> Byron, I have been having some success with R installation by replacing the fink version of g77 with the sourceforge version. So thanks very much for that info. Now that I have a good R base system compiled on OS X (I am trying both on a Powerbook for testing and on the XServe), I have been trying to load bioconductor to test compile with the sourceforge g77, but with no success over the last week. Something seems to have changed, since I was able to get further along before then. On the Powebook, I do the standard connect to http://www.bioconductor.org/getBioC.R and the do the getBioC(). It spins for a few seconds, then gives an error message "Error in getBioC()" but with no additional diagnostic information. This is the same if I ask for default or specific packages. I also tried "verbose = TRUE" and got no additional info. When I try on the XServe, I get an error message "Error in file(file, "wb") : unable to open connection". I am sure that I wasn't getting these messages when I first tried loading bioconductor to the XServe. Any idea of what is up? Thanks Mike Redmond UW-Madison ---
Byron Ellis wrote:
On Nov 2, 2004, at 2:11 PM, Michael Redmond wrote:
I have a working version of R 2.0 with option packages installed on an OS X 10.3.5 system. Option packages were installed with install.binaries option and that seemed to work. Install using sources failed. I tried installing bioconductor from sources and got mixed results (at best). Some packages failed to install (annotate, graph, Rgraphviz and maybe XML and possibly others). I also need to install arrayCGH to get DNAcopy. That uses the g77 compiler and failes with a mismatch of gcc library versions. I have g77 from fink.
You want the g77 from http://hpc.sourceforge.net---the fink g77 has never worked properly for me (actually, thats a general statement about fink for me :-) )
Is there a install.binary option for bioconductor and some of the bioconductor contributed packages (specifically arrayCGH)? If not now, when (if ever). If never, I may need some help to fix compile errors trying from source.
Jan De Leeuw usually wraps up the latest Bioconductor release into his "Batteries Included" version of R. I don't think there's a dedicated OS X machine doing binary builds anymore (IIRC, there used to be one doing the nightly smoke test). Unfortunately, OS X is probably the most finicky platform at the moment as well for that reason.
I am working on this for an Apple 4 node XServe to see if it can be used under the iNquiry portal for R/Bioconductor applications.
Ooh! Could you ping me off list about how thats working out for you? Our group has been considering one to serve some of internal computational needs---we're mostly curious about how much administration and configuration it requires. (i.e., do you just turn it on and go?)
Thanks Mike Redmond UW-Madison Statistics
_______________________________________________ R-SIG-Mac mailing list R-SIG-Mac@stat.math.ethz.ch https://stat.ethz.ch/mailman/listinfo/r-sig-mac
--- Byron Ellis (ellis@stat.harvard.edu) "Oook" -- The Librarian
_______________________________________________ R-SIG-Mac mailing list R-SIG-Mac@stat.math.ethz.ch https://stat.ethz.ch/mailman/listinfo/r-sig-mac