[R] R Install problems with Redhat 9 : "tcl.so" versus "tcl.so.0"

[Peter Dalgaard BSA]

"M.Kondrin" <mkondrin at hppi.troitsk.ru> writes:

> > [root at localhost root]# locate libtcl
> > /usr/lib/libtcl.so
> > /usr/lib/libtcl8.3.so
> > /usr/lib/libtclstub8.3.a
> > Is the "libtcl.so" really the "libtcl.so.0" that R is looking for ?
> > If so,
> > do I try and fool R somehow ?

We had a correspondent from Estonia (Allan Sims) run into that exact
problem (look in the archives). Last I heard, it didn't quite work to
just make the symlinks as suggested below -- rpm would still complain,
even with the links in place. What might work is to "rpm -i --force",
but there could be further problems down the road.

> > Any idea why libtcl.so.0 is not being found in my installed
> > tcl-8.3.5-88 ?

Probably a deliberate cleanup action from RedHat. By their convention,
they can break binary compatibility, including things like that, when
bumping the major version number.

> > If this is heading towards source code compiling, please be
> > explicit, I've
> > not done this before !
> > Grovelling for advice,

I suspect that Martyn will have made RPMs of 1.7.0 for both RH8 and 9
before you'd get through...

Actually, that's not quite fair. Building from source only takes half
an hour *if you have all the prerequisites*, but you lose the package
management (R is very easily uninstalled from /usr/local/, though).
Also note that RH9 has a compiler glitch that needs working around. I
think that was in the thread with Allan Sims too.

> If you haven't /usr/lib/libtcl.so.0 on your system I would suggest to
> make a symlink ln -s /usr/lib/libtcl8.3.so /usr/lib/libtcl.so.0
> 
> May be ldconfig is also required.
> 
> May be you will need to repeat this operation with libtk.so.0

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907