Date: Tue, 23 Apr 1996 14:04:52 -0400 (EDT)
From: dunder <dunder AT nyongwa DOT montreal DOT qc DOT ca>
To: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
cc: djgpp AT delorie DOT com
Subject: Re: V2 = hmmm
In-Reply-To: <Pine.SUN.3.91.960423084614.18939E-100000@is>
Message-ID: <Pine.LNX.3.91.960423135047.16701A-100000@noether.nyongwa.montreal.qc.ca>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII




On Tue, 23 Apr 1996, Eli Zaretskii wrote:

> You don't have to install the library sources.  Even if you need to 
> consult the sources occasionally, you can keep it as a ZIP file and look 
> inside with any of the archive browsers that are available.
> 
> The library in v2 is larger, that's true, but it's because it includes 
> much more functionality than v1.x did.  And btw, the library sources in 
> v1.x also required many MBytes to be installed.

Yeah but v2 has a lot more files :) And each files take 8k at creation! 
Anyhow, thanx for the advice about the ZIP file :)

> Please post the smallest source file that triggers these messages when 
> compiled.

I did actually, but I had an editing problem and I guess it got lost.. so 
here is the part which used ScreenPrimary ..

  if(option_nb < 5)
  {
   int pos_y = (option_nb * 3) + 4;
   gotoxy(1,1);
        ScreenPrimary[pos_y * 80 + 31] =
          ((ScreenPrimary[pos_y * 80 + 31] & (0x00FF)) | 0x7C00);
   gotoxy(30, pos_y);
   textattr(120);

Making this error .. the lines 136 and 137 point to the two lines of 
ScreenPrimary above. The lines 155 & 156 are pointing to the same kind of 
structure ... (it worked with v1.12)

rooedit.cc:136: invalid types `long unsigned int[int]' for array subscript
rooedit.cc:137: invalid types `long unsigned int[int]' for array subscript
rooedit.cc:155: invalid types `long unsigned int[int]' for array subscript
rooedit.cc:156: invalid types `long unsigned int[int]' for array subscript


I also had to add a #include "c:\djgpp\src\libc\co80\sc.h" in order for 
GCC to be able to actually recognise the variable ScreenPrimary (which 
now point to some other structure..)

> The only other way to run 32-bit protected-mode programs on top of 16-bit
> real-mode DOS is to write a DOS extender.  The extender takes much more 
> memory and has more compatibility problems, as DJGPP v1 has shown.

Well it didn't have to have a DPMI host. Although DJGPP may come with 
one, writing any commercial, or even shareware programs that require a 
special driver is somehow tricky. Salvation for Windows users but DOS 
users will bark!

> And my experience surely doesn't suggest that v2 needs much more memory. 
> Please try to use the advice of the FAQ on system setup; there must be
> something wrong with that if you cannot compile programs on a 8MB machine. 

Yep, as stated I would, I read the FAQ. But the problem was with 
Smartdrive which ate 2.5M of my memory ;) 
 
> What does go32-v2 print when invoked with no arguments in the DOS box?  
> My guess is your DPMI memory settings don't give enough memory to DPMI 
> clients.  Check the property sheet of the DOS box, under ``Memory''.  The 
> DPMI memory should be set to 65535K, not ``Automatic'', to let DJGPP 
> programs use as much virtual memory as your system can have.

DPMI memory available: 5420 Kb
DPMI swap space available: 2556 Kb

and I setted the DOS Box to 16M which as far as I could set.

thanx... try to help with the code :P