DM> The problem _seems_ to be something about the global 
 DM> variables I have, or at least that's all I can figure 
 DM> out.  The only reason I doubt this is that a DLL loaded 
 DM> by seperate processes should have its own global data 
 DM> space... shouldn't it?

 DM> If so, then nothing should conflict between the two 
 DM> processes (instances of the command interpreter, 
 DM> whether I use 4OS/2 2.52A or the Warp CMD.EXE).

 DM> If not, would I get a Sys3175 error when using a 
 DM> pointer that was created in the first process ... in a 
 DM> second one?  Or how about a semaphore created in the 
 DM> first process?

You really figure things out better AFTER you ask a question, even within a
few minutes after sending that.  

Alright, new question.  I have the "solution".  However, I use
the IDE, and, well, can't figure out where to put this solution.  I hacked
the makefile to get the test version working.

Here's the stats:

        Watcom 10.6 for OS/2
        Need the wlink line to have:

                wlink d all SYS os2v2 dll initi op m exp
                                          ^^^^^

However, I can't find a place to put that "initi" in the IDE. 
(BTW, there is one other option required, "op many", but I found
a radio button for it in the 4th screen of options under the linker.)  Any
directions to something I've obviously missed (or at least HOPE I've
obviously missed ) would be appreciated.  I'm also going to email
Watcom, but I'm not entirely sure I'll get a response from them. 
  I want to release this DLL before my classes start to kill
me.  :-)

(As an aside, for those who may be worried, the global variables are
protected by an _anonymous, private_ semaphore, which is what I wanted -
because the data should be _anonymous, and private_.)

--- Maximus/2 3.01