ML> I suspect that this is because Open Watcom is more strict on type
 ML> conversion than previous versions of Watcom. There might be a
 ML> switch to relax it. Otherwise we'll have to edit the source to make
 ML> it work.

 ac> I can't think of a good reason why it shouldn't build 
 ac> with OpenWatcom so you're probably right about it just 
 ac> being more strict.  But before you start delving into 

I've read some of the Change Log from 11c -> 1.5 and they mention OW
being more ANSI compliant. I took a brief look at the OW and WC command
line options and couldn't find anything apparent.

 ac> the help files I'd probably try building it with an 
 ac> older version of OpenWatcom. You can download older 
 ac> versions of OpenWatcom from their FTP site.

 ac> ftp://ftp.openwatcom.org/archive/

Cool.

I was starting to wonder whether I could get old OW versions. Their website
isn't very intuitive (imo).

 ac> And it looks like Watcom 11.0c is there too, which I 
 ac> hadn't noticed before:

 ac> ftp://ftp.openwatcom.org/archive/11.0c/

That's great. I'm going to get myself a copy; because the one I have came
from a dodgy bittorrent. Heh.

 ac> So maybe try that first, then try progressively newer 
 ac> versions until it breaks.  :-)

Yeah. 

So far, I could only build "maxn.exe" and "maxnd.exe"
(the 32-bit NT executables) that work properly by first building 32-bit
SLIB with Watcom 10.6. Otherwise it complains about missing definitions.
I've tried all sorts of combinations, even exactly replicating all INCLUDE,
LIBS, WATCOM environments as they are for WC 11.0c and it still couldn't
build dv\s3\comdll. You have to use the INCLUDE, LIBS, WATCOM
>>>and<<< WC 11.0c wcc/wcc386.exe's for it to work :/ 

The DOS versions built without any problems (with WC 10.6).

--- Maximus/2 3.01