🚀 go-pugleaf

In article <exuptr.479.716145032@exu.ericsson.se> exuptr@exu.ericsson.se (exuptr@exu.ericsson.se) writes:
>
>>mcastle@cs.umr.edu (Michael R Castle) writes:
>>>I'm trying to use Turbo C++ 1.0 to compile the GNU swapping-spawn library.
>>>
>>>The library was originally designed for MSC 6.0 and uses a construct that 
>>>forces data to be stored in the code segment.
>>>
>>>Is there any similiar directive for Turbo??
>
>Maybe we can solve the origional problem.  How much data do you need to pass?
> ----------------------------------------------------------------------------
>    "This must be Thursday.  I never could get the hang of Thursdays"
It appears to be only about 1k of data (probably half that, but I didn't add
up the bytes).  Apparently everything in the text and data segments get swapped
out (some of code seg  too, I assume, or it defeats the purpose).  Just that 
forcing these pointers to the swapped memory make it easy to just swap the 
data and text segments without much worry.

I don't know any asm directives, so I don't know how to attack the problem
from the point of view of using a #define with asm directives to place the
location of the variables.

I could just compile to asm, and frob that to move location of the vars, but
I don't have TASM.  

The reason for all this is to try to figure out why the hell GNU-Make crashes 
on my system (I took everything out of config.sys/autoexec.bat, so no 
conflicts there).  I suppose I could just bypass the whole problem, and make
a library just does desqview calls that, instead of spawing in the same session,
starts up a new session to execute the command (just like gnu-make does on a 
unix system :-).  But I would like to solve to original problem, too.

Please, no suggestions to using another make or swapping library.  I've made my
decision to use these tools, and will stick by them, problems and all.

mrc