[Plplot-devel] Re: Tweaked ltdl.c -> new tarball

[Alan W. I. <ir...@be...>]

On Thu, 27 Mar 2003, Rafael Laboissiere wrote:

> Hi again,
>
> Could you please test the today's tarball in netbsd?  I put some
> fprintf(stderr) calls in ltdl.c in order to help us hunting down the SIGSEGV
> bug.

I am putting this back on list because the result is interesting.

The java tweak worked, and the showstopper error message is
now much more informative.

./get-drv-info echo xwin.la | sed 's/.la//' > xwin.rc
dlerror: Shared object "libX11.so.6" not found
libltdl error: can't open the module
*** Error code 1 (continuing)
all' not remade because of errors.

However LD_LIBRARY_PATH solves the problem!

export LD_LIBRARY_PATH=/usr/X11R6/lib/
[irwin@sparky drivers]$ ./get-drv-info xwin
xwin:X-Window (Xlib):1:xwin:5:xw

So we have two errors here.

(1) Something is wrong with the linking of either xwin.so or get-drv-info on
netbsd that is corrected by LD_LIBRARY_PATH.  When this segfault problem
first surfaced I reviewed this thoroughly for xwin.so, and I swear it is
right. But I haven't looked at the get-drv-info linking.  It is possible
that libtool has a netbsd bug here, and it does not support the hierarchical
linking model like it is supposed to.

Because of work pressures I won't be able to do any more until tonight.
However, tonight I will review both linking efforts in detail again and do
lots of netbsd linking experiments to try and sort this all out.  Also, lots
of running of examples with LD_LIBRARY_PATH set just to make sure everything
else is working properly on netbsd.

(2) Something is badly wrong with libltdl.  It should not segfault on
linking errors (dlopen which it wraps does not).  You have obviously worked
around the problems enough to get a decent report out of libltdl now, but I
hope you document the problem thoroughly and send in a bug report to libltdl
about it because the current unpatched behaviour of segfaulting on any
library error (you verified this happened in Linux as well if you simulated
a linking error) is completely unacceptable.

Alan
__________________________
Alan W. Irwin
email: ir...@be...
phone: 250-727-2902

Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the Canadian Centre for Climate Modelling and
Analysis (www.cccma.bc.ec.gc.ca) and the PLplot scientific plotting software
package (plplot.org).

__________________________

Linux-powered Science
__________________________