[comp.sys.apollo] C Math on DN10000

majka@moose.cs.ubc.ca (Marc Majka) (07/19/89)

Help!  I recently upgraded our DN10000 from 10.0 to 10.1, and
all the C math functions have disappeared.  For example:

    main()
    {
        float x, sin();
        x = sin(1.0);
    }

when compiled, gives the message:

    % cc foo.c
    undefined                       first referenced
     symbol                             in file
    matherr                             /usr/lib/crt0.o
    sin                                 foo.o
    ld warning: Output file a.out not executable

The "matherr" object also appears in a console message
when I reboot the node. I get something like:

    Loading Init.
    Undefined Global "ldexp" referenced in /lib/libc
    Undefined Global "ldexp" referenced in /lib/libc
    Undefined Global "matherr" referenced in /lib/libc

I've called the hotline about this.  The languages group
eventually gave up and declared that it was a system problem.
The system group has no idea what is going on.  I can't
believe that this is happening on every DN10000 which is 
running SR10.1.p!  The library files /lib/syslib and 
/lib/clib are there, but something must be missing.  Any
help would be greatly appreciated!  

---
Marc Majka  -  System Manager  UBC Computer Science

derstad@CIM-VAX.HONEYWELL.COM ("DAVE ERSTAD") (07/24/89)

>  all the C math functions have disappeared.

For what it's worth, we don't have this problem - your
sample program compiled and ran fine.  We are running 10.1.P.

The missing globals are all in syslib, not clib.  You might
do a "bind /lib/syslib -glob" and see what's there.  (Note
that is an Aegis /com/bind, not a Unix bind).

Dave Erstad
Principal Design Automation Engineer
Honeywell SSEC

pcc@apollo.COM (Peter Craine) (07/24/89)

In article <2512@ubc-cs.UUCP> majka@cs.ubc.ca (Marc Majka) writes:
>
>The "matherr" object also appears in a console message
>when I reboot the node. I get something like:
>
>    Loading Init.
>    Undefined Global "ldexp" referenced in /lib/libc
>    Undefined Global "ldexp" referenced in /lib/libc
>    Undefined Global "matherr" referenced in /lib/libc
>
>---
>Marc Majka  -  System Manager  UBC Computer Science

When you installed the 'c' compiler, it loaded a (now) bogus version
of /lib/syslib[probably.881].  Find syslib in your AA and copy it
into /lib.  Reboot, et voila, your problem should be history.


                        Peter Craine, NARC
[ BTW, I discovered this same thing when I brough my node to SR10.2,
  so keep this in mind for the future, folks.]

majka@moose.cs.ubc.ca (Marc Majka) (07/25/89)

Many thanks to all who have responded to my request for help with
the missing math functions on the DN10000.  The problem was solved
for me by a very helpful Software Engineer from Apollo Toronto.
It seemed that a beta C compiler I had tried out had clobbered
/lib/syslib.  This was know to be a problem with installing on a
SR10.1 node, but I had missed the warning message sent out with the
beta tape.  I will be executed by a firing squad for this, as soon
as I have finished documenting my code :-)

---
Marc Majka  -  Altar Boy of Apollo

ianh@merlin.bhpmrl.oz (Ian Hoyle) (07/28/89)

From article <2512@ubc-cs.UUCP>, by majka@moose.cs.ubc.ca (Marc Majka):
> Help!  I recently upgraded our DN10000 from 10.0 to 10.1, and
> all the C math functions have disappeared.  For example:
> 

[stuff deleted]

> I've called the hotline about this.  The languages group
> eventually gave up and declared that it was a system problem.
> The system group has no idea what is going on.  I can't
> believe that this is happening on every DN10000 which is 
> running SR10.1.p!  The library files /lib/syslib and 
> /lib/clib are there, but something must be missing.  Any
> help would be greatly appreciated!  

yep, it happened here to :-(

We had to reinstall from our AA, the 10.0 version of syslib. However
I'm informed that the system runs slower with this old syslib (any
comments out there ??) - it sure seems like it does.

Somebody has *really* goofed on this one  ...... I hope it all
gets properly rectified when we get (at long last) REAL compilers
for the 10000.

					ian

BTW has anyone seen the Omniback/8mm tape thingie for the 10000 yet ??
We are waiting for delivery on ours so I'm keen to find out about others
experiences with it.

-- 

                Ian Hoyle
     /\/\       Computer Systems Superintendent
    / / /\      BHP Melbourne Research Laboratories
   / / /  \     245 Wellington Rd, Mulgrave, 3170
  / / / /\ \    AUSTRALIA
  \ \/ / / /
   \  / / /     Phone   :  +61-3-560-7066
    \/\/\/      ACSnet  :  ianh@merlin.bhpmrl.oz.au
                Internet:  ianh%merlin.bhpmrl.oz.au@uunet.uu.net