Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!uwm.edu!uakari.primate.wisc.edu!aplcen!haven!adm!cmcl2!lanl!lambda!jlg
From: jlg@lambda.UUCP (Jim Giles)
Newsgroups: comp.lang.fortran
Subject: Re: FIRST public review - response fina
Message-ID: <14176@lambda.UUCP>
Date: 15 Dec 89 23:20:34 GMT
References: <50500174@uxe.cso.uiuc.edu>
Lines: 54

From article <50500174@uxe.cso.uiuc.edu>, by hirchert@uxe.cso.uiuc.edu:
> 
> 1. As a general rule, the committe tried to generate letters postdated by
>    several days in order to insure that respondents would have their full
>    15 days.  Apparently, something went wrong in your particular case, but
>    this shouldn't have been the norm and most certainly was not done
>    intentionally.

I did an informal survey of the people I know who received responses to
their first roung comments.  In all cases, the response letter was dated
_exactly_ ten days before the postmark on the envelope which carried
the responses.  Mine was the only one to actually be _mailed_ after the
close of the second review - but all of us had our 15-day reply interval
curtailed.

(Now, of course, a sample of 4 response letters in not adequate to
prove a systematic pattern.  But the exact ten day delay does seem
to be a bit too coincidental.)

> 3. I was not on the subcommittee that drafted the response to your comments on
>    pointers, so out of curiosity, I looked up your comment:
> 
>    "   Pointers are needed for many applications.  Although ALLOCATE and
>        DEALLOCATE serve most of the necessary functionality, linked lists
>        and other indirect referencing schemes are still left out.   "
> 
>    I know I certainly would read this as supporting the idea of adding a
>    pointer capability to Fortran.  What in this is supposed to tell the
>    committee that you wouldn't accept the particular way they chose to add
>    this capability?

What I was asking for was linked lists and "other indirect referencing
schemes" (like trees, DAGs, GRAPHS, etc.).  There is nothing in my comment
which even mentions using POINTERS as a poor (slow, inadequate) substitute
for RANGE/IDENTIFY.  The overwhelming majority of the current POINTER
proposal has nothing to do with the recursive data structures I asked
for.

You say: "I know I certainly would read this as supporting the idea of
adding a pointer capability to Fortran."  Well, so would I.  At the time
I wrote that I had no objections to adding pointers to Fortran.  You
know - POINTERS?  Those things that Pascal or Modula have?  Or, perhaps
the kind already found in many Fortran implementations (like Cray's)?
Even C style pointers (with all that ugly pointer arithmetic) would
have been preferable to what the committee chose to do.

However, my present objection (as regards the public review responses)
is the casual way that the response implies that the present form of
the pointer proposal was what I asked for!  As I pointed out before,
it rather like getting a MACK truck when I asked for an economy
car - and then being told that I was given what I requested.  I suppose
I _COULD_ use the truck to commute to work (just as I _could_ use
the present pointers for something useful), the rest of its abilities
are not only unneeded, but undesireable.