🚀 go-pugleaf

In article <4e8gde$57n@stc06.ctd.ornl.gov>
           kennel@msr.epm.ornl.gov "Kennel" writes:

>Robert Dewar (dewar@cs.nyu.edu) wrote:
>> Michael says (talking about overflow in C)
>
>> "An implementation must handle this and must document how it does so."
>
>> Yes, of course the documentation may say that the result is to delete
>> your system disk. Even if a more reasonable behavior is described, any
>> program relying on such an implementation dependent feature is of
>> course compromising its portability. One of the troubles with C is
>> that it is oh-so-easy to introduce implementation dependence.
>
>> It is interesting to note that a number of "bug" reports sent into gcc
>> are complaints that GCC does not duplicate some implementation
>> dependent behavior seen in some other compiler.
>
>This is typically taken as a fault in the people submitting the bug reports.

That is precisely what it is.

>I do not.  I'm sympathetic to 'ordinary' people, and harsh on computers and 
>their designers. 

You are probably looking for platform/OS standards. If these are inadequate
shout at the people who designed the platform/OS. If they are adequate
you have a strong case for pressuring the compiler implementor. But this
has nothing to do with the C language standard.

>So like Mr. Dewar, I see this as empirical evidence that C (and some ways 
>C++) permit and encourage far too much implementation dependence.

No, it means that the people reporting the bug simply don't know
the language well enough. C is defined well enough to do useful work,
these problems are usually down to poor coding. Where different
implementations act differently there is usually good reason to do so.

> Really,
>this *IS* a fault in the language design.  Just documenting it and saying
>"you're a bad dog for doing X Y and Z" does not fix the real source of the
>problem. 

One of the problems is the amount of poor teaching material out there.

>For C it was reasonably justified given the circumstances and age. 

C can stand on its objectives, not just its age.

>For a modern object-oriented applications development it's inexcusable. 

Do you mean C++? :-)

-- 
-----------------------------------------
Lawrence Kirby | fred@genesis.demon.co.uk
Wilts, England | 70734.126@compuserve.com
-----------------------------------------