On Thu, 4 Feb 1999 13:09:48 +0000 (GMT) Clive Page
<[log in to unmask]> wrote:
> On Thu, 4 Feb 1999, Ian Chivers wrote:
>
> > don't use language features that you don't need. how many people need
> > pointers to solve the problem they have?
>
> I find I need data stuctures a lot (after all, they replace COMMON blocks)
> and that nearly all of them have array components, and that they length is
> hardly ever known at compile time. In f90/f95 this requires a pointer
> component. I think an allocatable component is to be allowed in F2000,
> but that's no help at present.
>
i agree, but was replying in the context of the following/
out of curiosity were you solving the above with f77? i remember
trying to debug clever use of blank common when working with
the cdc kit at imperial college.
oh the joys of working on advisory.
:-)
>
> > We have interns in our department who are arriving without any
> > knowledge of fortran, and who are able to read and modify a
> > (well written) f77 program after one or two week, and to write
> > a program from scratch after one month. Is it still the case
> > with f90?
> >
if all you were doing was the same problem in f90 then
it wouldn't take much f90 to get up and running.
> > > redundant possibilies to disassociate them (NULL() and NULLIFY()).
this really caught me out when writing a simple linked list.
it complicates the code and a relatively simple algorithm gets
lost.
the sparse matrix routine jane coded is not as clean as
it could be. at the time of the book we were working with
f90 compilers and had to put up with what was available.
>
> Yes, isn't it now time to declare NULLIFY as "obsolescent" and to
> recommend its deletion from F3000 (or whatever comes next after F2000)?
>
> --
> Clive Page, e-mail: cgp (at) le (dot) ac (dot) uk
> Dept of Physics & Astronomy,
> University of Leicester.
>
----------------------
Ian Chivers
[log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|
