[Rd] Re: [R] Newbie questions

Peter Dalgaard BSA p.dalgaard@biostat.ku.dk
05 Nov 2000 01:43:49 +0100

Zsombor Cseres-Gergely <z.cseres-gergely@ucl.ac.uk> writes:

> There is something with the hurd ld.so, but I do not know, what. But I 
> 1) run 'make test' with --gui=none. It went up till dies with
> > library("eda") # returns invisibly
> Error in dyn.load(x, as.logical(local), as.logical(now)) : 
> 	unable to load shared library "/home/zs/R-1.1.1/library/eda/libs/eda.so":
>   /home/zs/R-1.1.1/library/eda/libs/eda.so: undefined symbol: R_rsort
> Error in library("eda") : .First.lib failed
> Execution halted

OK, so that shows that dynamic loading doesn't work.

> 3) sent a message to the Hurd folks. One reply was this:
> <quote on>
> The correct linker options for this sort of issue are the same as on Linux.
> Perhaps their build scripts behave differently on Linux than on Hurd.  It
> sounds like they expect to link the main executable with --export-dynamic
> (aka -E).  Use nm --dynamic or objdump --dynamic-syms to see what symbols
> are defined in the executable on Linux and on Hurd.
> <quote off>

Aha... Actually watching the link step during make would probably show
that -export-dynamic is not among the options. The codes that inserts
that flag during configure would seem to be (in configure or configure.in)


which likely will not trigger on the Hurd... In the longer run we
should fix configure, but on a short term, you can fix this by setting

MAIN_LDFLAGS = -export-dynamic

in Makeconf. (It would seem to be spelled with a single hyphen, for
some oddball reason).
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard@biostat.ku.dk)             FAX: (+45) 35327907
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