Re: Redhat 5.0 problems/solutions

[Paul E. Johnson]

On Sat, 6 Dec 1997, Marcus G. Daniels wrote:

> >>>>> "PJ" == Paul Johnson <address@hidden> writes:
> PJ> However, the rudest surprise was that the swarm package would not
> PJ> make. I kept seeing errors about mistakes in stdlib.h.  After a
> PJ> long time, I figured out that the make was looking first in the
> PJ> /swarm/include directory for header files, and those header files
> PJ> are "stale." they dont work with the libraries I have. SO, one by
> PJ> one, I renamed those libraries, such as stdlib.h, math.h, one of
> PJ> the ieee785.h.  Finally swarm made without errors because it found
> PJ> better header files. Note, I don't think you can just delete all
> PJ> of them because some are found nowhere else on the pc.
> 
> Hmm, what is /swarm/include?  For a given compiler target, there should be
> at most two versions of an include files, one in the usual place

I mean the /swarm-1.0.3/include directory that is created when
I untar the swarm source.  Don't you have that too???  If you
notice in the Makefile.conf, it is set to always look there for
include files first, before going to the otherincdirs you set
in the top of the Makefile.conf.  Because some of those header
files do not match the libraries as installed, Swarm won't make.
I've solved by renaming those header files and swarm makes. However,
I see now the easiest thing to do would have been to change the
Makefile.conf near the bottom so the include directories have
the swarm-1.0.3/include directory last, rather than first. It
wouldn't use any of those headers unless it needed to.

> I don't follow RedHat things, but I'd guess that RedHat 5.0 is based
> on the GNU C Library, like version 2.0.5 or something. 
They say it is GlibC. Is that meaningful to you?

Well, I think the really big problem is that none of the test
programs for libtclobjc have ever worked (recall make check doesn't
make, at least not for any linux user in recent memory).  Without
the test of libtclobjc, I don't know what the problem is, whether
it is due to Redhat 5 or if I just bungled the install of tclobjc.
I have been looking at it for a while and I think you
are right that the configure script for libtcl does not 
find the correct headers, and I'm working on it.  Incidentally,  
the blt demo programs work fine. 

Recall this was a "clean" install of redhat 5 and all swarm stuff, since
the redhat 5 "upgrade" script destroyed my redhat 4.2 install, so I
don't think I've got the lingering library problem.  

I was wondering, however, if there is any big reason why
I can't have a system that has tk/tcl 8.0 installed in the
Redhat places /usr/lib and /usr/include, 
and then the old tk/tcl in /usr/local/lib and /usr/local/include.
As long as I make swarm look there, perhaps by setting
environment variables TK_LIBRARY and TCL_LIBRARY, won't
it work?
Paul Johnson


                  ==================================
   Swarm-Support is for discussion of the technical details of the day
   to day usage of Swarm.  For list administration needs (esp.
   [un]subscribing), please send a message to <address@hidden>
   with "help" in the body of the message.
                  ==================================