D-Hertel@hq.asa.COM (Dave Hertel) (03/06/91)
Help! I'm trying to compile mmdf2-43 & MH6.7.1 on a DEC RISC based workstation 5000/200 running Ultrix v4.1 Both mmdf & MH compile ok (albeit w/misc. errors that the compiler seems content with ignoring) But 2 problems have surfaced: 1) inc & msgchk don't work. Both programs produce a segmentation fault (and sometimes a core dump) All other MH programs appear to function normally. During compiliation of the MH code, several times the error: "Warning: illegal pointer combination" shows up. I've tried compiling w/several different options set in the conf/MH file but alway have the same results. The compiler stop, is just give warnings.. 2) the SMTP daemon won't run. When the SMTP daemon is executed, it immediately quits with an error code "98". The network connection is operational as far as I can tell although I have not yet (but soon) connected the DEC to a ether backbone. Hints? Also, I've recently compilied this same code with same options on a SUN running ver:4.1.1. The SUN did not produce any of these problems. On the DEC, I tried to use the same options, except that I removed options that directly referred to SUNOS. Help & Hints would be appreciated. Dave Hertel
vossj@topaz.ucq.edu.au (03/08/91)
> I'm trying to compile mmdf2-43 & MH6.7.1 on a DEC > RISC based workstation 5000/200 running Ultrix v4.1 > Both mmdf & MH compile ok (albeit w/misc. errors that the > compiler seems content with ignoring) > > But 2 problems have surfaced: > 1) inc & msgchk don't work. Both programs produce a > segmentation fault (and sometimes a core dump) > All other MH programs appear to function normally. > > During compiliation of the MH code, several times the > error: "Warning: illegal pointer combination" shows up. > I've tried compiling w/several different options set in the > conf/MH file but alway have the same results. > The compiler stop, is just give warnings.. I've also been trying to get MH6.7.1 working on our DECstation 500/200 and I have exactly the same problem. I noticed that both inc and msgchk don't link properly, they both give the following error. ld: Undefined: macnum proxy macros macbuf Any thoughts, John ------------------------------------------------------------------------------ John Voss, InfoPSI: J.Voss@ucq.edu.au Computing Services, AARNet/Internet: J.Voss@ucq.edu.au University College of Central Queensland, Phone: +61 79 369804 Rockhampton, QLD. 4702 Fax: +61 79 361361 Australia. [DTE 050527372000077] -------------------------------------------------------------------------------
jch@hollie.rdg.dec.com (John Haxby) (03/08/91)
In article <1991Mar7.101306.6977@topaz.ucq.edu.au>, vossj@topaz.ucq.edu.au writes: |> I've also been trying to get MH6.7.1 working on our DECstation 500/200 and I |> have exactly the same problem. I noticed that both inc and msgchk don't |> link properly, they both give the following error. |> |> |> ld: |> Undefined: |> macnum |> proxy |> macros |> macbuf This is the source of your problems. The undefined symbols ensure that inc and msgchk don't produce fully resolved images--if you try to run something that has the relocation information in it it core dumps. There are two fixes for this; one is to wait for the next release of ULTRIX (which isn't pretty); the other is to extract ruserpass.o from /lib/libc.a on ULTRIX V4.0 and do the ar stuff to put it back into /lib/libc.a on 4.1. (These undefined symbols come from a dodgy ruserpass.o that slipped throug the net). If you don't have access to the V4.0 ruserpass.o; mail me directly and I'll send it to you. Equally, if you have the 4.3BSD ruserpass.c; you can compile it up and I believe it will work just fine. -- John Haxby, Definitively Wrong. Digital <jch@wessex.rdg.dec.com> Reading, England <...!ukc!wessex!jch>