Difficulties with egcs-1.0.3 on RedHat 5.2?

Peter Dalgaard BSA p.dalgaard@biostat.ku.dk
06 Mar 1999 01:26:58 +0100


Douglas Bates <bates@stat.wisc.edu> writes:

> Some machines in our department use RedHat 5.2 Linux but, to maintain
> consistency with other machines, the compilers as from egcs-1.0.3, as
> shown below.  When I compile R-0.63.3 on these systems I can't pass
> the tests in "make check".  The base-Ex.R file stops execution with
> the error shown below.
>  ...
... 
> This does not seem to happen on other Linux machines with other C and
> Fortran compilers but it does happen on a Solaris 2.6 for Intel
> machine with the same compilers.
> 
> Is this a known problem with egcs-1.0.3?

Yes, (only I thought it was in 1.0.2...). I reported it some months
ago. It's a register allocation foulup in the compiler, showing up
when inlining is used. If you *really* want me to, I can dig up the
reference...

Goes away if you compile without optimisation. Unless my memory fails
me, qbeta.c is the culprit, and you can get away with just recompiling
that one without -O. It's gone away in egcs 1.1.0, but that is a bit
of a massive download for the modem crowd.

-- 
   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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._