Re: [cc65] Chicken and Egg Problem (?)

From: Greg King <greg.king41verizon.net>
Date: 2009-08-24 18:05:17
From: "Oliver Schmidt"; on Sunday, August 23, 2009; at 04:01 AM -0400
>
> > From my perspective, it seems reasonable to presume that the linker,
> > together with the C/run-time library, does "somehow make sure" that
> > main() is referred.
> > I understand that the linker, for sure, isn't limited to C programs;
> > so, it doesn't make sense to hard-code such a reference.  But, what
> > about extending the linker config. syntax/functionality to allow
> > something analogous to .forceimport (being kind of the "opposite" of
> > SYMBOLS {}, maybe called IMPORT {}).  The built-in linker configs.
> > then could use that functionality for the symbol _main.
>
> Just imagine forgetting or misspelling main().  Wouldn't you expect the
> linker (called with -t <target>) to complain about not being able to
> resolve the symbol '_main'?

2) No, because the built-in configure-code _is_ "hard-wiring".  We do not
want the "-t" option to force people to stuff an artificial "_main:" label
into their purely assembly-coded programs.  But, it is a nice idea for
custom configure-files.

1) Look at the FORMAT section of "lunix.cfg"; it shows a hint of this idea.
And, the beginning of "atari.cfg" shows the export part of this idea.

Instead of IMPORT {}, an "import" attribute can be added to the SYMBOLS
section.  It would be the opposite of the "value" attribute.
SYMBOLS {
    __STACKSIZE__: value = $800;
    _main: import;
}

----------------------------------------------------------------------
To unsubscribe from the list send mail to majordomo@musoftware.de with
the string "unsubscribe cc65" in the body(!) of the mail.
Received on Mon Aug 24 19:14:05 2009

This archive was generated by hypermail 2.1.8 : 2009-08-24 19:14:08 CEST