To: boylesgj AT lion DOT cs DOT latrobe DOT edu DOT au
Cc: djgpp AT delorie DOT com
Subject: Re: DJGPP vs Borland C++.
Message-ID: <19970131.153607.4511.0.fwec@juno.com>
References: <5caj0a$7kd AT lion DOT cs DOT latrobe DOT edu DOT au>
From: fwec AT juno DOT com (Mark T Logan)
Date: Fri, 31 Jan 1997 15:30:40 EST


On 24 Jan 1997 15:03:38 GMT boylesgj AT lion DOT cs DOT latrobe DOT edu DOT au (Gregary J
Boyles) writes:
>I have found Borland C++ and Turbo C++ to be very buggy and there is 
>no
>way I would part with several hundred dollars for such unreliable
>products. DJGPP is far supperior to any Borland products, at least as 
>far
>as DOS compilers go.
>
>I have encountered many problems with Borland products as well as many 
>of
>the lectures and tutors at Latrobe University. The problems don't seem 
>to
>surface until you try to write large and complicated programs with 
>them.
>Some of the problems are as follows.
>
>1) The IDE and PC will nearly always hang at some point when stepping
>through a program with the integrated debugger.
>

Are you making sure that _ALL_ TSR's are not active when you run the
integrated
debugger?  This is stressed in the manual.

>2) Despite the fact that all optimizations are turned off the 
>integrated
>debugger will skip lines etc, the watch window will state that 
>variables
>are undefined when this is clearly not the case and the watch window 
>will
>state that a variable is optimized and not available when the 
>execution
>bar contains the variable. It makes it close to impossible to debug
>anything!
>


>3) The IDE freezes regularly on key presses, until you hit another 
>key,
>and you end up with two characters instead of just one.
>
>4) The Turbo Debugger for DOS provides no way of tiling windows with a
>single key press or mouse click as you can with the IDE.
>
>5) Borland C++ 4.5 for Windows provides no way of changing the target
>executable etc of a project except when you are creating a project (as 
>far
>as I can see).
>
>6) The exe files they produce will nearly always hang at some point in
>there execution and it seems to be often on entry to or exit from
>functions which indicates that the stack is not being managed 
>properley.
>
Or perhaps you let an unitialized pointer corrupt the stack?
TC++ is _not_ protected mode, i.e. there is no memory protection


>7) One exceptionally bizare problem was that one of my programs worked
>perfectly when run from the hard drive but would hang when run from a
>floppy (again on entry to a function).
>
>I'm sure half these problems stem from that rediculous memory model 
>system
>and near and far pointers etc they have used.
>

I will side with you here, many of my programming cronies have 
complained about programs with  far pointers freezing

>What is also frustrating is that there is virtually nil run time error
>support with programs just hanging the PC without even a hint as to 
>what
>went wrong such as 'segmentation fault'. Occasionally you may get 
>'stack
>over flow' however half the time this seems to have nothing to do with
>what went wrong.
>

I think your criticism of TC is not entirely warranted, but all the same
I too would take DJGPP over TC without a second thought.

-Fwec