Date:	Mon, 9 Nov 92 09:29:26 EST
From:	DJ Delorie <dj AT athena DOT ctron DOT com>
To:	mcastle AT mcs213k DOT cs DOT umr DOT edu
Cc:	djgpp AT sun DOT soe DOT clarkson DOT edu
Subject:	exceptioon 6

> (aside: what is the memory complexity of compilation, or is it
>NP?)

It depends on the complexity of the largest single function.
Complexity is measured by loops, branches, and statements.  -O nearly
doubles the amount of memory used by compile.

>As slight possibility is that memory gets corrupted swapping back and forth,
>and something gets munged (though I would think this unlikely in that go32
>always reads code back in from the a.out file, right?)

go32 reads unmodified pages from the a.out file instead of swapping.
Stack may get toasted, though, and cause an invalid return.

>unless, when swapping a page table, the pt gets munged and decides to point 
>into data somewhere :->

That could happen, but it's much less common and thus much more unlikely.

>Anyway, what I had to do to get around this was strip _EVERYTHING_ out of
>memory to give go32 all I could.  To compile the one file took 45 minutes.
>I used debug32 instead of go32, but I'm not sure if this solved the problem
>(I wouldn't think it would make a difference), but it's neat to watch the 
>little number across the top... gives you something to do while you harddrive
>thrashes. :->

Change go32's makefile to always define TOPLINEINFO=1 and you'll get
that line all the time.  I do it myself, and set the "mono" flag in
go32 so that line shows up on the mono monitor.

>I haven't tried compiling the file with 1.09 (too scared to :-), so I don't
>know if the problems still there.

I don't know of any changes that would have fixed whatever the problem
is.

If anyone *can* figure this one out, that person would be a hero - the
generic problem has been around a while and nobody has been able to
fix it yet.

DJ
dj AT ctron DOT com
Life is a banana.

- Raw text -