| Phillip Helbig---remove CLOTHES to reply 2004-11-17, 3:59 pm |
| In article <m1mzxgz9x0.fsf@MLMCE0000L22801.local>, Richard E Maine
<nospam@see.signature> writes:
>
> That is *WAY* *WAY* outside of the bounds of anything in the scope
> of the standard.
I realise that, and that the implementation would probably be quite
difficult. On the other hand, IIRC a compiler is required to say that a
program is too complex for it if that is the case; from a user's point
of view this is something similar. I was just wondering if folks
thought it would be a good idea. :-|
I suppose it is a quality-of-implementation issue.
> Remember that the standard doesn't even have the concept of a compiler
> at all.
Right, it's always "the processor".
> So there's a *LOT* of fundamentals to cover here. And to get
> into issues of versioning of compilers? Hmm... Would that include
> standardizing versions that didn't conform to the standard so that
> once you ever had a compiler bug, no future version could ever be
> standard-conforming because standard-conformance also implied
> standard-conformance of previous versions?
This is more than I need. Presumably, the constructor of the processor
knows, or can suspect, if something in the format of the results of
compilation has changed, and could write the compiler version into those
results, to be checked by whatever version of the compiler wants to USE
those results.
> All my opinion, of course, even if I expres it somewhat vigorously.
No problem; your opinion and expertise are respected here!
(By the way, apologies for any confusion caused by me posting my
original problem in comp.lang.fortran and on the comp-fortran-90 mailing
list. I was thinking that (were it in fact some subtlety I had missed)
that the latter might have more experts, while in the former I might be
more likely to find someone with the same or a similar compiler with
whom to compare results.)
|