home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.lang.c++
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!destroyer!ais.org!draper
- From: draper@ais.org (Patrick Draper)
- Subject: Re: Death of Borland C++ Compiler
- Message-ID: <BzH1LF.9yA@ais.org>
- Organization: UMCC
- References: <chaolin.88@fcit-m1.fcit.monash.edu.au>
- Date: Fri, 18 Dec 1992 20:15:13 GMT
- Lines: 51
-
- In article <chaolin.88@fcit-m1.fcit.monash.edu.au> chaolin@fcit-m1.fcit.monash.edu.au (CHAOLIN CHANG) writes:
- >bundled. But as my programs are becoming complex (using pointer, pointer
- >to pointer, pointer to pointer to pointer and farmalloc extensively),
- >Borland compiler begins to fall over. One of my colleagues is also
- >suffering for the same experience when he tried to push the Borland compiler
- >to its limits.
- >
- >I had pulled all my hair out. If the program is run within the debuger line
- >by line, it works fine. But as soon as I let go, it hung my PC.
- >
- >I recompile the same source code in Zortech. It WORKS.
- >
- >I wonder if anyone else has the same experiences with me?
- >
- >Charles
- >18/12/92
- >
-
- The Borland compiler is reliable for programs of almost any complexity. I
- imagine there are some limits of loop nesting, etc, but if you reach thouse
- you almost certainly are doing something wrong.
-
- In your case, it sounds like you've got memory bugs. The fact that you're
- using pointer->pointer->pointer type expressions indicates to me a hard to
- debug program. The compiler shouldn't have any trouble with these, I've
- never had these problems at all.
-
- Try using the 80386 debugger. It loads the program into the same memory
- address as DOS does from the command line. Any location dependent errors
- will show up as they do without the debugger.
-
- The program when compiled with Zortech also will have the same bugs as those
- existant under Borland, the difference might be segment ordering (different
- program arrangement) under the different compiler. Bugs that clobber unused
- memory are still bugs, even if the effects are not obvious.
-
- By the way, I DO know what I'm talking about because my current project at
- work is a Point of Sale package that is over 70,000 lines long and requires
- a DOS extender to run. It is compiled under Borland C++ 2.0 and 3.1.
-
- ---One more thing, you might want to try the DOS extender included with
- Zortech because it will help show your memory bugs. Development with a DOS
- extender is much much better than plain old unextended DOS.
-
-
- ------------------////////////////////\\\\\\\\\\\\\\\\\\\\------------------
- | Patrick Draper-ZBT Disclaimer: I can't control my fingers, |
- | draper@umcc.ais.org I can't control my toes! - Ramones |
- | University of Michigan Computer Club |
- NO CARRIER We are a nation of laws, not people |
- ------------------\\\\\\\\\\\\\\\\\\\\////////////////////------------------
-