In a message dated 12-2895, Jorgen De Gier said to Mike Bilow:

 MB> On the BC++ 2.0 for OS/2 CD-ROM, there is what looks like a second
 MB> of the compiler.  What is it and why is it there?

 MB> 11-10-94   2:00a   1087499           0  BCC.EXE
 MB> 11-10-94   2:00a    791052           0  BCC2.EXE

>bcc2.exe is full 32bit and no thinking. I just read this in a newsgroup in
>a message from Nick Knight. He used to visit this echo, maybe he can give
>you more details if you give him a yell.

Yes, I'm here :)

Bcc2 is exactly as you describe it.  It is not documented all that well,
and I'm not sure exactly why.

bcc2.exe, in general, produces, hmmm, how do I say this tactfully
 ... more stable code than bcc.exe.  But, it doesn't thunk. 
AND, it produces a few weird problems on it's own.  The things it screws up
are more livable, the biggest being a "compilation module not
found" message when trying to debug into a module relatively deep into
an object module list.  To "remedy" this, you reorder your linker
script so that the modules needed are closer to the "start". 
Very inconvenient.

The consensus here is to use bcc2.exe for everything you can.  Relink if
debugging problems occur.  If you need thunking, try to localize it and use
bcc.exe for those part(s), bcc2.exe for the rest.

Nick

.. To whom the gods destroy, they first teach Windows... -- MR/2 2.26 #0

--- InterPCB 1.50