Re: Big array indexes

helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply) w
rites:

> In article <m18y8xwgyd.fsf@MLMCE0000L22801.local>, Richard E Maine
> <nospam@see.signature> writes:
> 
>
> It seems to me that adding suffixes to constants in order to indicate
> the KIND makes the code less readable.

I agree (as I mentioned in the material right before the extract.

> Would it be too much to ask to  have something like
>
>    DEFAULT_INTEGER_KIND = SELECTED_INT_KIND(15)

Well, I seriously doubt you would get that exact syntax accepted,
but that's a trivial detail. Viable Syntax can be crafted and
isn't particularly difficult.

Things along this line have been proposed multiple times. It was
even on the accepted feature list at one time for either f95 or
f2003. (It was long enough ago that I get  as to which one.)
But its proponent had thought it was going to be easy to do. Turned
out to be far more work than he had anticipated, and it didn't get
done.

I think I might have also seen it on a list of proposals being developed
for f2003+.  That was (much) more recent, but I don't recall for
sure. If it was on, I don't think it got adequate support to stay on
and get further development of the proposal.  (F2003+ is still really
in the stage of developing proposals, with not a single vote having
yet been taken at the WG5 level, or even much discussion there, so
all that part is pretty early - still at the stage of developing
proposals in a bit more depth.)

Anyway, the poooint of the above is that

1. Yes, the basic idea has lots of atttractions to lots of people.

but

2. It turns out to be a pretty big job.  I haven't detailed exactly why
it is so big.  I could say some things of the top of my head, but it
would probably tuen longer than I want to write right now, and
I'd forget about a bunch of the less obvious traps.

People have tried to find ways to "cheat" and avoid a lot of the
rewriting all over the standard, but every attempt so far has
not really held together. I'm afraid it just needs rewriting all
over the standard... and then there is the implementation
work.

So maybe it will happen some day. A fair amount of weight on both
sides of the scale.  A lot of desire for it vs a lot of work to do it.
So far, the "lot of work" side has outweighed the other, but that
could change.

--
Richard Maine
email: my last name at domain
domain: summertriangle dot net