Xref: utzoo comp.unix.ultrix:7134 comp.mail.elm:4090
Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!think.com!spool.mu.edu!uunet!bonnie.concordia.ca!ccu.umanitoba.ca!silver.cs.umanitoba.ca!gedetil
From: gedetil@silver.cs.umanitoba.ca (Gilbert E. Detillieux)
Newsgroups: comp.unix.ultrix,comp.mail.elm
Subject: Re: Anyone ever run elm on DECstation/Ultrix?
Message-ID: <1991May7.230339.3520@ccu.umanitoba.ca>
Date: 7 May 91 23:03:39 GMT
Article-I.D.: ccu.1991May7.230339.3520
References: <1991May06.023124.1453@cs.fau.edu> <1991May6.213607.14679@unixland.uucp>
Sender: news@ccu.umanitoba.ca
Organization: Dept. of Computer Science, University of Manitoba, Winnipeg, CANADA
Lines: 52

In article <1991May6.213607.14679@unixland.uucp>, bill@unixland.uucp (Bill Heiser) writes:
> In article <1991May06.023124.1453@cs.fau.edu> liq@shark.cs.fau.edu (Qiang Li) writes:
> >I am trying to install elm on our new machine DecStation 5000 200CX, most
> >likely run Ultrix 4.1, but I heard someone had trouble installing it on
> >their DecStation 3100 and elm may not support Ultrix.  Is this true?
> 
> I just installed elm 2.3 PL 11 on a bunch of DECstation 5000/200's 
> and 3100's.  I didn't have any trouble installing it, and haven't seen
> any problems with using it so far.  I havne't used it much, though, so
> your mileage may vary.

I've been running elm 2.3 PL6 for several months, and just installed PL11 last
week, on a DECstation 5000 and two DS2100's running Ultrix 4.1.

The only problem I had with the installation was trying to compile the
thing with Gnu C, rather than the standard C compiler:  the problem seems to be
with inconsistencies in the Ultrix header files (e.g. typedef int time_t; was
in some headers, rather than using long int).  When I gave up on Gnu C's nice
strict type-checking and used the much more lax MIPS C compiler, everything
worked fine.  (Retorical FAQ:  When will DEC get Unix right?)

I did run into a slight problem with the VAX Ultrix 4.1 compilation of the
src/signals.c file, due to the bizarre handling of the "void" type in its
C compiler, but the following patch took care of that:

*** signals.c.dist      Thu Aug  9 17:41:10 1990
--- signals.c   Thu May  2 10:23:17 1991
***************
*** 30,36 ****
  #include <signal.h>

  #ifdef        VOIDSIG
! typedef       void    sighan_type;
  #else
  typedef       int     sighan_type;
  #endif
--- 30,38 ----
  #include <signal.h>

  #ifdef        VOIDSIG
! /*typedef     void    sighan_type;
! */
! #define sighan_type   void
  #else
  typedef       int     sighan_type;
  #endif

-- 
Gilbert E. Detillieux			Internet: <gedetil@cs.umanitoba.ca>
Dept. of Computer Science		Phone: (204)474-8161 or 8313(messages)
University of Manitoba, Winnipeg MB	Fax:   (204)269-9178
CANADA  R3T 2N2				"Build bridges, not walls"