🚀 go-pugleaf

On Tuesday, in article
     <1995Nov21.100102.1@gst-soft.demon.co.uk>
     cwood@gst-soft.demon.co.uk "Chris Wood" wrote:

CW> I am using -Ff=1024, which solved this problem when I put in the Netware

Probably tells the compiler to put any lumps of data over 1024
bytes into far data -- slightly slower and bulkier code for refs to
such data.

CW> mods; but don't really understand what it does or if I can change it

Try reducing the number, e.g. halving it, to get more
moved out of the 64K.

CW> again to solve this problem. Failing this I presume that I can turn off 
CW> some of the #defines in config.c to not include certain servers, thereby 
CW> saving some valuable space in the data segment, or not?

.c or .h?  Yes, you can same some space that way.  That
was the kind of thing I did to produce the 'tiny' version
of TextWin (compare the source, if you want to know the
things I kicked out, but perhaps they're things you want).
You are already compiling the 'small' version of ka9q,
I hope?

--
John Washington

In article <817046728.24098@thokk.demon.co.uk>,
Malcolm Muir <Malcolm@demon.net> wrote:

>Indeed, if you were to just use -Ff all variables become far.

All the Borland compilers I've used treat "-Ff" as being "-Ff=32767".

You may have to go down to -Ff=63 or 31 in order to get KA9Q to
link, depending on what options you have configured in.


-- 
Andrew Benham       A.D.S.Benham@bnr.co.uk
BNR Europe Ltd, 140 Greenway, Harlow Business Park, Harlow,
   Essex CM19 5QD, United Kingdom
Tel: +44 1279 402372    Fax: +44 1279 405746

In article <1995Nov21.100102.1@gst-soft.demon.co.uk>
           cwood@gst-soft.demon.co.uk "Chris Wood" writes:

> 
> I have now got some pop3 sources for ka9q, which when I compile in are 
> okay, but during linking I get a DGROUP > 64K error message. I only use 
> the Borland compiler for building this and have little understanding of 
> the switches.
> 

I've been looking at producing a 32-bit version using the DJGPP compiler.[*]
however I don't understand what the myriad of command-line switches mean

CFLAGS= -w -w-par -a -d -f- -A- -G- -O -Z -DMSDOS -Ff=1024

 -O == optimize ?
 -D == define MSDOS
 -Ff == far handling [not needed]

I suspect some of the others modify alignment of data within structs, would
anyone like to provide a decode of the switches.

the flag /C is passed to lib, I assume this means create a library.

RjL
[*] /pub/simtel/vendors/djgpp in ftp.demon.co.uk -- the gcc compiler & tools
+----------------------------------+
|    richard@illuin.demon.co.uk    |  ... and the light of the Lamps of the
+----------------------------------+      Valar flowed out over the Earth

Andrew Benham (adsb@bnr.co.uk) wrote:
> In article <817046728.24098@thokk.demon.co.uk>,
> Malcolm Muir <Malcolm@demon.net> wrote:

> >Indeed, if you were to just use -Ff all variables become far.

> All the Borland compilers I've used treat "-Ff" as being "-Ff=32767".

I have never used the -Ff form so have no experience. I prefer to set
things explicitly then there is no ambiguity :)

Looking at the manual I have for 3.1 it is unclear. In one place it
defines -Ff as 'create variables as far'. in another is defines it as
'create variable over default size as far, default being 32767'. Now I
know why I like to be explicit :)

-- 
Malcolm S. Muir
Sunderland
England

Chris Croughton (chris@keris.demon.co.uk) wrote:
> In article <817324091snz@illuin.demon.co.uk>
>            richard@illuin.demon.co.uk "Richard Letts" writes:

> >I've been looking at producing a 32-bit version using the DJGPP compiler.[*]
> >however I don't understand what the myriad of command-line switches mean

I suspect you are in for a disappointment, I don't recall the reason
now, but we looked at using GCC for ka9q some time ago and it was not
possible. Could it have been that there is no support for access to
the real mode interrupt routines ?

-- 
Malcolm S. Muir
Sunderland
England

In article <817324091snz@illuin.demon.co.uk>
           richard@illuin.demon.co.uk "Richard Letts" writes:

[snip]

> I've been looking at producing a 32-bit version using the DJGPP compiler.[*]
> however I don't understand what the myriad of command-line switches mean
> 
> CFLAGS= -w -w-par -a -d -f- -A- -G- -O -Z -DMSDOS -Ff=1024
>
>  -O == optimize ?
>  -D == define MSDOS
>  -Ff == far handling [not needed]
> 
> I suspect some of the others modify alignment of data within structs, would
> anyone like to provide a decode of the switches.
> 
> the flag /C is passed to lib, I assume this means create a library.

One of the things I find horrible about PD software is the lack of
documentation. A particular example of this is the failure to document
switches in makefiles.

I infer that you do not understand the switches used in a makefile which
expects a Borland compiler and which you want to port to DJGPP. I got
the following from Borland C++ 3.1 User's Guide, Chapter 5: The
command-line compiler.
-w        Display warnings on
-w-par    Suppress -par warning messages: Parameter <name> is never used
-a        Align word
-d        Merge duplicate string on
-f-       Don't do floating point
-A-       Use Borland C++ keywords
-G-       Select code for speed
-O        Removes jumps to jumps, unreachable code, and unnecessary
          jumps
-Z        Enable register load suppression optimisation
-DMSDOS   Define MSDOS to the null string
-Ff=1024  Create far variables automatically; sets the threshold to 1024

In Borland C++ 3.1 Tools and Utilities Guide, Chapter 3: The tlib /C
option is used "to add a module to a library that includes a symbol
differing only in case from one already in the library"
----------
walter briscoe

In article <817324091snz@illuin.demon.co.uk>
           richard@illuin.demon.co.uk "Richard Letts" writes:

>I've been looking at producing a 32-bit version using the DJGPP compiler.[*]
>however I don't understand what the myriad of command-line switches mean
>
>CFLAGS= -w -w-par -a -d -f- -A- -G- -O -Z -DMSDOS -Ff=1024

-w      Turn warnings on
-w-par  Turn off specific warning (unreferenced parameters)
-a      align on word boundary
-d      merge duplicate strings
-f-     floating point off
-A      ANSI, no extensions
-G-     Optimise for size
-O      Optimise jumps
-Z      Suppress register reloads (a specific optimisation)
-DXXX   define MSDOS for preprocessor
-Ff=n   Put variables over n bytes long into separate segments

For your purposes you can ignore everything except -D and the word
alignment...

>the flag /C is passed to lib, I assume this means create a library.

Case sensitive library (not the default).  Again, you can ignore it (the
GCC library and linker follow Unix standards, I believe).

***********************************************************************
*  chris@keris.demon.co.uk      *                                     *
*  chriscr@cix.compulink.co.uk  *  FIAWOL (Filking Is A Way Of Life)  *
*  100014.3217@compuserve.com   *                                     *
***********************************************************************