home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.os2.programmer
- Path: sparky!uunet!caen!batcomputer!msiadmin.cit.cornell.edu!bai
- From: bai@msiadmin.cit.cornell.edu (Dov Bai-MSI Visitor)
- Subject: Re: 2.0 fails miserably in the PARANOIA program ( was: Division by 0 crashes OS2/2.0 )
- Message-ID: <1992Sep13.210932.322@tc.cornell.edu>
- Sender: news@tc.cornell.edu
- Nntp-Posting-Host: msiadmin.cit.cornell.edu
- Organization: /usr/local/lib/news/organization
- References: <gyan.716247010@unixg.ubc.ca> <1992Sep11.225553.5638@tc.cornell.edu> <92Sep12.023824.23909@acs.ucalgary.ca>
- Date: Sun, 13 Sep 1992 21:09:32 GMT
- Lines: 30
-
- In article <92Sep12.023824.23909@acs.ucalgary.ca> bauwens@acs.ucalgary.ca (Luc Bauwens) writes:
-
- >I reran the dpara check under OS/2 2.0 on my machine with those flags
- >-FPi87 -G2 -AL -Od, and surprise surprise, MS Fortran 5.1 came out
- >with only one SERIOUS DEFECT (after milestone 160), and completed the
- >test.
- >
- >But, BTW, for 1./0., the answer was, as before, 1. And interestingly,
- >Paranoia did not flag that as being a problem.
- >
- >(Dov: I could uuencode my dpara.exe and e-mail it to you. That would
- >let us establish whether the difference we note is between 5.0 and
- >5.1 or between your and my machine.)
- >
- >Luc
- >
-
- Well, I checked the executable of Luc (thanks Luc!) on my machine
- and it gave the same results as my executable: It crashed 2.0, but
- 1.3 passes with 1 FLAW. I could not check under DOS, since the executable
- was not bound. I tried to bind it with my libraries, but it failed
- to run under DOS with a runtime error.
-
- So, this establishes that the difference in not caused by the difference
- in compiler versions.
-
- Dov
-
-
-
-