John Venier writes:
> On the other hand I was not programming in 1977, so things could
> well have been very different then.
Yes. Things do change over 25 years. Quite a lot in the computer
business. Enough that some things that are obvious now were arcane
or impractical then. The environment really was quite different,
and some big differences aren't necessarily things you'd think of
at first. Not having text editors, for example - it does make a
huge difference.
I've had people criticize a computer selection I helped make in
the early 80's (quite a bit more recent than the '77 standard),
saying we were fools for not having selected one of vendors x, y,
or z instead (particular names purposely omitted). Apparently
they weren't aware that none of the vendors they suggested actually
existed at the time (well, I think one of them might have been
a new-start that nobody had yet heard of and that didn't yet have
any actual offerings).
Time does matter. The environment changes. Plus we learn.
Often the learning is from mistakes made in the past. I like to
explain that one of the ways I got so good at uncovering Fortran bugs
was that I've made all of those mistakes. (Which isn't nearly so
embarassing as admitting how many times I've made the same ones).
> I'm perhaps also in the minority as I am constantly trying to update
> Fortran 77 code, and in my opinion a big improvement I can make to
> such code is to replace the old idiosyncratic ways of getting machine
> constants with the Fortran 90 intrinsics. Unfortunately the fact that
> the old ways were idiosyncratic means there is not a straightforward
> way to do this, but it does give me many opportunities to reflect on
> the Fortran 77 standard.
Sounds to me like you are being complementary. You are saying that
you see the f90 standard as a big improvement. One can phrase that
as saying that the f77 standard was in such a shape that a lot of
improvement was possible. I'd agree with both statements as being
true, but I prefer the tone of the second one.
The reflection, in any case, does seem worthwhile - learning from
history and all that stuff.
Now if you want to look at something that really needed improvement,
the f66 standard provides plenty of material. I was going to add a
"war story" here, but I think I should do some real work instead.
--
Richard Maine
[log in to unmask]
|