Date: Mon, 1 May 1995 09:04:35 -0400
From: kagel AT quasar DOT bloomberg DOT com
To: bdavidson AT ra DOT isisnet DOT com
Cc: djgpp AT sun DOT soe DOT clarkson DOT edu
Subject: Re: your mail RE: mine on Strange problem.
Reply-To: kagel AT ts1 DOT bloomberg DOT com

   Date: Mon, 1 May 1995 00:23:42 -0300 (ADT)
   From: Bill Davidson <bdavidson AT ra DOT isisnet DOT com>

   Oops, no subject header.  This is about your problem with your sort 
   routine, (not)getting the attention of your pager when invoked with a 
   disk spec in argv[0].

   I have no experieence using popen(), but your intriguing message brought 
   a couple of questions to mind.
   1) How many logical or physical drives do you have?  Is, for example, d: 
   a valid drive spec on your machine?  If so, what drive is usually current 
   when you are working?

I have 2 physical drives each with 2 compressed logical drives and, obviously,
one small uncompressed 'drive'.  C: and H: are compressed and live on the C:
physical drive which DOS has re-mapped to E:.  D: and F: are compressed and
live on the D: physical drive which is mapped to I:.  (I don't know what DOS
has against the G: drive letter.)  I was attached to the C: logical drive and
the executeables live there.

   2) I have no idea what effect this might have, if any, but are your DOS 
   TEMP and GO32TMP variables pointing to the same directory, or different 
   ones?

I'll have to check on this one.  My TEMP and TMP env. vars point to D:\copy,
DJGPP lives on D:\djgpp and I **think** I put the GO32TMP under that, but I
will have to check.

   It certainly sounds as though something is using that "c:" to override a 
   default, hence my wondering about available valid drive specifiers.  I 
   would aslo check all my environment variables that point at directories, 
   and ensure that they start with a drive spec, not just "\tmp" or 
   something. 
   It's a fascinating problem, keep me (us) posted.
   Bill

I also had another problem a few month's ago which I never followed up on and
which may be related.  I have an enhanced version of pr which I originally
wrote for the Datalight "C" compiler (of blessed memory, later Zortech C++, now
Symantec C++) and long ago ported to UNIX for it's features, one of which is
writing directly to the spooler (lp or lpr) on UNIX or to PRN: on DOS.  On
porting it back to DOS/DJGPP I noticed that writing to LPT1:/PRN: would
furiously churn the disk but produce nothing!  I have taken to writing the
output to disk and then printing that from DOS.

-- 
Art S. Kagel, kagel AT ts1 DOT bloomberg DOT com