[comp.emacs] Unexec of GNU on SYS V

alan@gsi2.UUCP (Alan Whitton) (07/07/87)

HELP,

GNU seems to work fine on a Plexus (a 68020 based SYS V box), except for
unexec, is there anyone who has any hints which might be causing it to
bomb out?

		thanks   alan  8-)

Karl.Kleinpaste@cbstr1.UUCP (07/13/87)

alan@gsi2.UUCP writes:
   GNU seems to work fine on a Plexus (a 68020 based SYS V box), except for
   unexec, is there anyone who has any hints which might be causing it to
   bomb out?

Having just looked at src/m-plexus.h, I see that NO_REMAP is
explicitly turned off for the Plexus box.  Try turning it on and
remaking.  The result will (should) be that unexec() will almost
certainly generate a valid xemacs, but that the text/data boundaries
will not have been shifted around.  The vagaries of COFF are not to be
believed.  The non-shift of text/data boundaries merely means that a
couple hundred Kb worth of data space won't be shared amongst the
(potentially many) simultaneous incarnations of emacs.

Karl

Karl.Kleinpaste@cbstr1.UUCP (07/13/87)

Oh, BY THE WAY...

alan@gsi2.UUCP writes:
   GNU seems to work fine on a Plexus (a 68020 based SYS V box), except for
   unexec, is there anyone who has any hints which might be causing it to
   bomb out?

*Which* GNU?  *Which* Plexus box?  *Which* revision of the OS?
*Which* system configuration, if it might possibly be significant?

For crying out loud, I keep pretty up-to-date on SysVisms of GNU
Emacs, and I tend to get a lot of mail from people who've noticed that
fact, asking questions about this SysVish OS and that 3B processor.
(And RLK handles questions about X peculiarities, and bob@osu-eddie or
I get questions about UUCPing GNU things from osu-eddie, and I don't
know who handles most ordinary BSD questions...)  Last week, I got
mail from someone who asked me some generic question about why GNU
Emacs wouldn't build on his SysV VAX.  After about 4 messages back and
forth, lo and behold, the bozo was in possession of (gag) GNU Emacs
16.something.  Let's just say that 16.anything didn't have a
snowball's chance of ever really running on a SysV box of any flavor.

If you're going to ask questions of the whole blasted planet, would
you PLEASEPRETTYPLEASE...
    [a] Tell us *exactly* which version of GNU Emacs you've got.
	Don't say "18.xx," because the "xx" matters one heck of a
	lot if you're on a 3B processor, as a case in point.
    [b] Tell us *exactly* which processor you're using.  "VAX" is
	not necessarily sufficient; MicroVAXen have been known to
	have problems not found in an 11/780.  "3B" isn't sufficient,
	either; a 3B15 is completely different from a 3B20.
    [c] Tell us *exactly* which operating system you've got.  Just
	saying that you've got "a VAX" is hardly sufficient, since
	that doesn't even tell us if you're running 4BSD or SysV.
	"VAX running BSD" isn't sufficient, either, since you might
	be running (e.g.) 4.1BSD, which is going to require an
	answer quite different from 4.3BSD.  Similarly for SysV.0
	versus SysV.2.
    [d] Tell us anything that might possibly be odd about your
	particular situation, especially if it's on a system/OS
	combination which is supposed to be readily supported.  Do
	you have a newer C compiler than the generic one which your
	configuration would normally have?  Is your SysV not *quite*
	SysV (e.g., has Berkeley IPC been grafted into it?)?  Did
	your vendor supply an unusual signal library?

Sorry to be appearing (in all probability) quite flammable; it's not
really intended that way.  But trying to answer questions without
knowing the whole situation can be mighty frustrating.

Cheers,
Karl