[BioC] [R-SIG-Mac] Mac os x graphviz-Rgraphviz problem

Mon Feb 5 10:32:00 CET 2007

Just add one thing to what Kasper already said: I suspect that the issue
is that the  .so from graphviz are not visible to the loader.  You can fix
up the LD_LIBRARY_PATH in someway: "make install" from graphviz has some
msgs like

=============================================================
PATH="$PATH:/sbin" ldconfig -n /home/vital-it/lilong/bin/graphviz-2.12/lib
----------------------------------------------------------------------
Libraries have been installed in:
   /home/vital-it/lilong/bin/graphviz-2.12/lib
......
==============================================================

that hints how to do it.  I didn't see "ldconfig" quoted in the msg...

Li

> Your example works fine for me on Mac OS X ppc 10.4.8 with R-2.4.1
> and graphviz 2.12 compiled from source.
>
> You can check your graphviz version from R by doing
> R> graphvizVersion()
>
> You have an error in the Graphviz version reported below, I assume
> you are talking about Rgraphviz' version. When you run it in the X11
> xterm (which I assume you are talking about), it seems as if some
> PATH/library settings are messed up. I don't know anything about
> Darwinports configuration, so I am afraid I cannot help much.
>
> Getting Rgraphviz to work requires some connection between graphviz
> and Rgraphviz, and it is important for us to know what you installed,
> and especially how. I assume Darwinports compile from source? Since
> you have installed several different flavours of Graphviz, are you
> certain that you have cleaned up after you? Using the Graphviz binary
> provided by Simon Urbanek should work, but I find it  bit puzzling
> which versions (and how you installed them) you are discussing below.
>
> So, try to reinstall Rgraphviz the way you describe below. Note what
> version of Graphviz R is linking to, both by looking at the compile
> statements when installing the package (which is why you need to
> reinstall) as well as trying
> R> graphvizVersion()
> and then see if something is not matching.
>
> Btw. you are most certainly not running Graphviz 2.8 under MS
> Windows, unless you have taken major pains to do so.
>
> Personally, Graphviz used to be a pain to install on the Mac, but the
> new version (2.12) compiled easily for me.
>
> Kasper
>
>
>
> On Feb 1, 2007, at 1:45 PM, Gavai, Anand wrote:
>
>>
>> Hello everyone,
>> I am having problems making Rgraphviz working with graphviz.
>> I installed R from Darwin ports "http://darwinports.opendarwin.org/
>> getdp/"
>>
>> by
>>> sudo port install r
>>
>> I set my environment variables in my .bash_profile e.g export PATH=/
>> opt/local/bin:$PATH
>>
>> I also installed other packages which are required for Rgraphviz
>> like "graphviz" and "pkgconfig" by
>>
>>> sudo port install graphviz
>>
>>> sudo port install pkgconig
>>
>> both these commands will also install dependencies which are
>> required. After this
>>
>> I start up an R session being root and get the latest version of
>> bioconductor packages by sourcing it from bioconductor website
>>
>>> source("http://www.bioconductor.org/biocLite.R")
>>> biocLite()
>>
>> waited for a while and then install
>>
>>> biocLite("Rgraphviz",depend=TRUE,type="source")
>>
>>
>> it worked fine when i opened it in a Terminal window, only the
>> loading part but when I run something like this
>>
>>> library("Rgraphviz")
>>> a<-c(1,0)
>>> b<-c(1,1)
>>> d<-rbind(a,b)
>>> graphObj = ftM2graphNEL(d)
>>> plot(graphObj)
>> it gives me the following error messages
>>
>> *** caught bus error ***
>> address 0x40, cause 'non-existent physical address'
>>
>> Traceback:
>>  1: .Call("Rgraphviz_doLayout", graph, as.integer(type), PACKAGE =
>> "Rgraphviz")
>>  2: layoutGraph(g)
>>  3: agopen(x, name = name, layout = TRUE, layoutType = y, attrs =
>> attrs,     nodeAttrs = nodeAttrs, edgeAttrs = edgeAttrs, subGList =
>> subGList,     recipEdges = recipEdges)
>>  4: .local(x, y, ...)
>>  5: plot(graphObj)
>>  6: plot(graphObj)
>>
>> Possible actions:
>> 1: abort (with core dump)
>> 2: normal R exit
>> 3: exit R without saving workspace
>> 4: exit R saving workspace
>> Selection:
>>
>> I cannot put here the sessionInfo() as i need to abort it, the
>>
>> the Graphviz version i have is 1.310.  I also tried to install and
>> configure 2.8 previously with no luck.
>>
>>
>> in  X11
>> when i load this library("Rgraphviz") in X11 interface it gives
>> following error message
>>
>> Error in dyn.load(x, as.logical(local), as.logical(now)) :
>>         unable to load shared library '/Library/Frameworks/
>> R.framework/Resources/library/Rgraphviz/libs/i386/Rgraphviz.so':
>>   dlopen(/Library/Frameworks/R.framework/Resources/library/
>> Rgraphviz/libs/i386/Rgraphviz.so, 6): Symbol not found: _agfstnode
>>   Referenced from: /Library/Frameworks/R.framework/Resources/
>> library/Rgraphviz/libs/i386/Rgraphviz.so
>>   Expected in: dynamic lookup
>> Error: .onLoad failed in 'loadNamespace' for 'Rgraphviz'
>> Error: package/namespace load failed for 'Rgraphviz'
>>
>> the sessionInfo() here is
>>
>>> sessionInfo()
>> R version 2.4.1 (2006-12-18)
>> i386-apple-darwin8.8.1
>>
>> locale:
>> C
>>
>> attached base packages:
>> [1] "tools"     "stats"     "graphics"  "grDevices" "utils"
>> "datasets"
>> [7] "methods"   "base"
>>
>> other attached packages:
>> geneplotter    annotate     Biobase       graph
>>    "1.12.0"    "1.12.1"    "1.12.2"    "1.12.0"
>>
>> Any help in this regard is appreciated. I had simimar problem in
>> the past with Suse Linux we managed to solve it, and my Windows
>> station is working fine with Graphviz 2.8 and Rgraphviz
>>
>> only my mac system (intel tiger) is a bit of a problem. I manually
>> tried to install graphviz and Rgraphviz using the source files and
>> setting LD_LIBRARY_PATH and PKG_LIBRARY_PATH but in vain the only
>> way to install them together without errors is through a darwinport.
>>
>>
>> Regards,
>> Anand
>>
>> _______________________________________________
>> R-SIG-Mac mailing list
>> R-SIG-Mac at stat.math.ethz.ch
>> https://stat.ethz.ch/mailman/listinfo/r-sig-mac
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives:
> http://news.gmane.org/gmane.science.biology.informatics.conductor
>