From: "Sammy Mitchell" 
@Date: Wed, 22 Jan 2003 15:33:07 -0500
@Sender: semware-owner{at}sawasdi.apana.org.au



Carlo Hogeveen wrote:
>No, then the compiler generates an error. It is a choice
>again. Semware was of the opinion, that it was better to
>forbid duplicate definitions, then to let one override the
>other. In hindsight I would prefer the last definition to
>override previous ones, and let the compiler generate only
>a warning.

I'll of course consider this.  I guess we'd have to add an
intermediate code generation phase (we're strictly single
pass right now) with abstract addresses, so that we could
remove redefined code, and then generate the final code with
actual addresses (well, as absolute as you can get for a
Virtual Machine).

>In hindsight again, the same would have
>benifited key definitions too, which we can currently
>redifine in the same macro, but in the wrong order: now we
>have to put in extra forward declarations.

Strange but true, the key redefinition 'feature' was sort of
an accident.  It was one of those things that sort of
unintentionally worked the way it does now, and instead of
removing it, we left it in.

>It is an imperfect world,

Definitely!

>let's reboot it.

The world, or TSE?  Or both?

I'm not sure - are you saying you want to "give us the
boot"?

--
Sammy Mitchell

--
TSEPro mailing list



---
[sawasdi.apana.org.au] (3:800/846.13)