Gabriel Dos Reis wrote:
> root <[EMAIL PROTECTED]> writes:
>
> | do a symbolic link from libreadline.so.4 to libreadline.so
>
> All suggestions I've seen so far look like that one. Can anyone
> explain why that is supposed to do anything sensible?
>
> I would not like to see Axiom, a scientific computational platform,
> built on superstition.
We should really ask Martion what the following commands give:
ldd /users/rubey/axiom/mnt/linux/bin/AXIOMsys
ldd /users/rubey/axiom/obj/linux/bin/interpsys
I ask about interpsys because it is used during build and normally
it also links readline. On Debian systems there is
'/usr/lib/libreadline.so' which is a link to '/lib/libreadline.so.5'
AXIOMsys compiled on machine where libreadline.so.5 is installed
should look for libreadline.so.5 (and not for libreadline.so.4).
So either Martin is using binary build on different system or
there is somewhere libreadline.so.4 which was picked when linking
AXIOMsys, but is not in the paths searched by dynamic linker.
One typical case is when program contains its own sources of
shared libraries -- program may succesfully link against such
shared library but it can not run because by default dynamic
linker searches only standard locations (in fact I suspected
for a moment that Axiom has own copy of readline stashed
somewhere, but after checking I think it is not the case).
I would try first typing:
locate libreadline.so.4
which on one of Debian machines found '/lib/libreadline.so.4'
(this one is from compat package and should work fine, but
something more strange may cause troubles).
--
Waldek Hebisch
[EMAIL PROTECTED]
_______________________________________________
Axiom-developer mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/axiom-developer
