1. RE: Secrete new routine
The real problem with no forward referencing is that your program ends up upside
down.
Having the main routine at the very bottom of the program makes its readability
quite poor. Neither can you organize your procedures where similar ones are
together. It's quite valuable to place the uninteresting routines out of the way
at the bottom of the program. It's also irritating to have the organization of
your program controlled by an arbitrarily enforced rule. I had always assumed
that
this limitation was because it was a "single pass" compiler. But now obviously
it
is not the case.
2. RE: Secrete new routine
- Posted by "Kat" <gertie at visionsix.com>
Dec 13, 2004
-
Last edited Dec 14, 2004
On 13 Dec 2004, at 7:54, George Walters wrote:
>
>
> posted by: George Walters <gwalters at sc.rr.com>
>
>
> The real problem with no forward referencing is that your program ends up
> upside
> down. Having the main routine at the very bottom of the program makes its
> readability quite poor. Neither can you organize your procedures where similar
> ones are together. It's quite valuable to place the uninteresting routines out
> of the way at the bottom of the program. It's also irritating to have the
> organization of your program controlled by an arbitrarily enforced rule. I had
> always assumed that this limitation was because it was a "single pass"
> compiler.
> But now obviously it is not the case.
It never was the case. Derek or DavidC or someone (besides me) pointed out
that while the forward reference raised an error, the compiler didn't need to
crash there. It could log all the possible targets, then process the error list
against the target list at the end of compiling. Throw away the resolved
errors, anything remaining is a real error. Presto, no problem.
Kat
