From: muta...@gmail.com   
      
   On Monday, August 9, 2021 at 10:43:42 PM UTC+10, Joe Monk wrote:   
   > > > > Some people said   
   > > > > the same thing about GCC exceeding 16 MiB.   
   > > > > They insisted that I should rewrite GCC to make   
   > > > > it use disk files, rather than changing the underlying   
   > > > > machine and OS to suit my purposes. We went in   
   > > > > different directions, and only one direction has a   
   > > > > GCC that works.   
      
   > Now thats not true.   
      
   Not sure what your definition of "true" is, as you seemed   
   to completely acknowledge everything I said.   
      
   > The simple truth is that no one (so far)   
   > has wanted to invest the time necessary to take a super   
   > old copy of GCC and make it run on MVS3.8J in less than   
   > 16MB,   
      
   Isn't that what I just said?   
      
   > when there are superior products (like JCC and   
   > Dignus) available.   
      
   If that is the motive other people have for doing   
   diddly squat, so be it.   
      
   It's not my motive, and I don't agree with the premise.   
      
   > It wouldnt be too difficult to work with GCC 3.2.3   
   > to make it run in a region of around 3MB using disk   
   > work files.   
      
   Nothing is too difficult for the man who doesn't have   
   to do it himself.   
      
   Personally I find it difficult to even get GCC 3.2.3 to   
   stop getting negative indexes. My heart was is my   
   mouth when I tried this change:   
      
   C:\devel\gcc\gcc\config\i370>cvs diff -r 1.51 -r 1.52 i370.h   
   Index: i370.h   
   ===================================================================   
   RCS file: c:\cvsroot/gcc/gcc/config/i370/i370.h,v   
   retrieving revision 1.51   
   retrieving revision 1.52   
   diff -r1.51 -r1.52   
   1156c1156,1162   
   < #define Pmode SImode   
   ---   
   > /* Theoretically using DImode instead of SImode should stop   
   >    the negative indexes being generated, which are a problem   
   >    if we run a 32-bit executable while in AMODE64, but there   
   >    seems to be something else broken in the i370 target preventing   
   >    that from working */   
   > #define Pmode DImode   
   > /* #define Pmode SImode */   
      
   But it actually made things worse.   
      
   And besides that, what is the problem being solved?   
   Who is trying to compile (with GCCMVS) C programs   
   so large that it exceeds the limits of S/370? I personally   
   do that (and not on my own source code), but is anyone else?   
      
   > I have actually looked into this, but just havent had the   
   > time to sit down and make it happen.   
      
   Yeah, it's amazing just how much fantastic vaporware there   
   is out there bottlenecked on that exact thing.   
      
   BFN. Paul.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)