DN>
  > JdeBP> The problem that I've been having, is that (try as I might), I canno
  > JdeBP> find a single file handle leak in the program that invokes NMAKE (I
  > JdeBP> even went to the lengths of using OPEN_FLAGS_NOINHERIT on *all* call
  > JdeBP> to DosOpen in the program).
  >
  > Is there any REXX involved? A lot of REXX execs open files and then
  > don't close them with a call to STREAM(...,'C','CLOSE').
DN>

  No REXX at all, I'm afraid.  It's all straight C and C++.

DN>
  > JdeBP> And until now I found it difficult to believe that the presence of
  > JdeBP> Netware Client for OS/2 or Communications Manager/2 (both of which
  > JdeBP> were the differing factors in tests) could affect file handle usage.
  >
  > I had a number of problems with the Netware requestor, but that was on
  > a machine I used primarily for mainframe development work. I never ran
  > out of file handles, but I hit a number of other problems. It wouldn't
  > surprise me in the least if that were at least part of the problem.
DN>

  I'm a little leery of laying the blame at the fault of the Netware
  Client because, as I said, it was just one differing factor.  In fact,
  most of the machines where this problem has manifested itself don't
  have the Netware Client for OS/2 on them at all.

  I was just wondering whether this sort of file handle abuse was
  endemic in device drivers.

  And, incidentally, I've been off work sick, so haven't been able to
  test out whether there are differing file handle limits on different
  machines yet.  I'll let you know when I can.

  > JdeBP <
___
 X MegaMail 2.10 #0:

--- Maximus/2 3.00