[R-SIG-Mac] Problems installing igraph
steeves at raingods.net
Mon Oct 28 20:29:12 CET 2013
On 24.10.2013 22:42, Simon Urbanek wrote:
> we don't support macports so I cannot provide more than a general
> First, your output appears to deal with a dirty package - I don't see
> any compilation output and yet you have a dylib - that doesn't seem
> right. Make sure you're building from a clean tar ball (and *not* a
> directory!) - you have cut out the important part so we can't tell.
I'd snipped the compilation, as it was pretty lengthy and didn't
contain anything that stood out to me.
I've attached a text file with the output from the compilation, on the
off chance that something in it helps.
> If you are sure that you're building off a clean tar ball then my
> guess would be that there is some problem with the linking flags used
> in igraph. R build process doesn't perform link-time symbol
> resolution, so errors in flags are not apparent at link time - they
> only show up at run-time. Typically, to trace issues like this,
> search you libraries for the missing symbol and add that library to
> the link flags. nm is your friend when checking the symbols.
I downloaded R.app from CRAN, and the prebuilt igraph looks like it
installed without complaint. Looking at that igraph.so with nm, I don't
see any reference to that function
15:23 steevmi1 at PHUSCA-L24849
% nm ./igraph.so | egrep '_colamd'
000000000003c570 T __glp_colamd
000000000003c0f0 T __glp_colamd_recommended
000000000003c550 T __glp_colamd_report
000000000003c2c0 T __glp_colamd_set_defaults
Google shows some references to this function in SuiteSparse, so I'm
uninstalling it and will try to install igraph again, and see if maybe
there was some sort of conflict there, but any other thoughts you might
have would be greatly appreciated.
Michael Steeves (steeves at raingods.net)
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
More information about the R-SIG-Mac