[Info-vax] x86 Compiler Update

[invalid]

On 2018-07-31, Dave Froble <davef at tsoft-inc.com> wrote:
> On 7/31/2018 4:14 PM, invalid wrote:
>> On 2018-07-29, Scott Dorsey <kludge at panix.com> wrote:
>>> invalid  <address at is.invalid> wrote:
>>>> On 2018-07-25, Bob Koehler <koehler at eisner.nospam.decuserve.org> wrote:
>>>>>
>>>>>     Fortran 77 seemd mostly to be a growth from FORTRAN IV.  But Fortran
>>>>>     90 introduced a different language that only vaguely resembles it's
>>>>>     predecessors.
>>>>>
>>>>>     My initial reaction to Fortran 90, was "nice language, but it's not
>>>>>     Fortran".
>>>>
>>>> Agreed. This is the reaction of pretty much everybody who spent any time
>>>> coding FORTRAN. I do like Fortran also, it adds some nice features, but
>>>> still falls short of what it should be.
>>>
>>> I like the matrix manipulation stuff, which makes it MUCH easier for a
>>> vectorizing compiler to work, and also makes matrix code a lot easier
>>> to read.  It was long overdue.
>>>
>>> But there are new features like pointers that make it much easier for
>>> engineers to shoot themselves in the foot.
>>
>> So what? Fortran's pointers are still much safer than most. And I think they
>> don't go nearly far enough. As it is, you have to use C for certain things
>> because Fortran pointer support just isn't what it should be.
>>
>>> The huge advantage of f77
>>> was that it was much more difficult to totally screw things up than it
>>> was with C, Ada, or PL/1.  The lack of pointers and dynamic memory
>>> allocation went very far toward improving code reliability.
>>
>> I think you will find it very difficult to screw anything up with Fortran
>> pointers.
>>
>>>
>
> I think that with access to pointers, it's rather easy to screw things 
> up.  I like having that access, but, one must know one's limitations.

That's pointers in the abstract. Fortran pointers are implemented pretty
safely, to the point of not being very useful. Like I said, you will have to
work hard to screw things up.

If you want to debate, try writing a piece of code where you can screw
anything up and when you can't, let's talk about it some more ;)