Re: [TCLCORE] Tk_MainEx: extreme archeology

[Alexandre F. <ale...@gm...>]

On Wed, Dec 10, 2008 at 6:08 PM, Donald G Porter <dg...@ni...> wrote:
>
> Is there anyone who can advise me on the intended purpose of the
> routine Tk_MainEx() ?  In particular, why it is in the public stubs
> table?  What callers are expected to call it through the table?
> Do such callers exist?
>
> The routine was added in Revision 1.3 of tkMain.c with log message:
>
> revision 1.3
> date: 1999/03/10 07:04:42;  author: stanton;  state: Exp;  lines: +34 -4
> integrated stubs into 8.0 main branch
>
> Some understanding of this may be relevant to my getting thread
> safety issues right in the TIP 338 implementation.


Looking at the comment in the 'changes' file:

 3/10/99 (new feature) Tk now uses the new stub library feature in Tcl.
 The Tk library now contains no direct references to any symbols in
 Tcl.  In addition, there is a new Tk_MainEx() function that takes an
 interpreter as an argument.  See the Tcl documentation for more
 information about the stubs mechanism. (redman)

it looks like there was no intention of special handling; Tk_MainEx
got stubbed just like dozens of others.
Now since it is stubbed, it becomes a macro via tkDecls.h. So any
caller (like a wish-like app) including the proper headers must be
calling it through the stub mechanism. Even those just calling Tk_Main
which is itself a macro calling Tk_MainEx.

I feel like I'm stating the obvious. What am I missing in your question ?

-Alex