Path: utzoo!censor!geac!torsqnt!hybrid!scifi!bywater!uunet!spool2.mu.edu!sdd.hp.com!elroy.jpl.nasa.gov!ncar!csn!ub!dsinc!netnews.upenn.edu!vax1.cc.lehigh.edu!lusgr
From: lusgr@vax1.cc.lehigh.edu (Steve Roseman)
Newsgroups: comp.unix.aix
Subject: Re: NEW GNU Emacs available - with DUMPING!!!
Message-ID: <183.279d57ce@vax1.cc.lehigh.edu>
Date: 23 Jan 91 14:30:53 GMT
References: <1133@troi.dbaccess.com> <182.279c284e@vax1.cc.lehigh.edu> <14622@june.cs.washington.edu>
Organization: Lehigh University
Lines: 42

In article <14622@june.cs.washington.edu>, slh@wolf.cs.washington.edu (Scott Heyano) writes:
> In article <182.279c284e@vax1.cc.lehigh.edu> lusgr@vax1.cc.lehigh.edu (Steve Roseman) writes:
> [stuff]
> |Emacs version 18.56, now available from prep.ai.mit.edu has full RS/6000 
> |support, including DUMPING!!!!!!  Simply get /u/emacs/emacs-18.56.tar.Z, do
> |the normal installation stuff (like making paths.h and config.h) and it works.
> |(use m-ibmrs6000.h and s-aix3-1.h in config.h, configure X-Windows if desired).
> |
> 	Are you sure dumping is working?
> 	The MACHINE file clearly states that you can produce a dump,
> 	but after some period of time the dumped executable will fail.
> 	Quote:
>   Dumping does not work.  Code has been written to implement it, but
>   it fails because the address of bss seems to vary occasionally
>   between Emacs runs.  It does not seem to vary from minute to minute,
>   but every few days or weeks it changes to a new steady state.
>   When this happens, the dumped Emacs data file becomes invalid.
-- 

The instabilities that existed seem to have been completely resolved.
Earlier, (pre-mid-December), it was weird - a dumped copy would start and stop
working depending on rebooting, configuration changes, room temperature, etc.
Eg. A copy of Emacs would work until a reboot, a new one would be made, the
second would work for a few hours, then suddenly, the second was dead, but the
first worked again. 

However, for the past month, I have NO PROBLEMS at all (except an Emacs
compiled at AIX 3002 wouldn't work on a 3001 system, for a different reason.)

Is it coincidence that a week after I sent Richard Stallman the newest code to
test, that 18.56 is released, with the same code?  I suspect the MACHINES file
was simply not updated, or the disclaimer left in, just in case. 

If anyone has problems, please feel free to contact me.

Steve

-------------------------------------------------------------------------
Steve Roseman
Lehigh University Computing Center
LUSGR@VAX1.CC.Lehigh.EDU