Re: [TCLCORE] tkMain.c requires tclInt.h, why?

[Donald G P. <dg...@ni...>]

David Gravereaux wrote:
> The only source file that requires Tcl's internal interface is tkMain.c and
> only for the two functions:
> 
> Tcl_GetStartupScript
> Tcl_SetStartupScript
> 
>>From the naming convention, they look to slated for moving to the public
> interface.  Could they be moved before Tcl 8.5 goes final? 

That was the intent.  Just hasn't gotten done yet.

A few partial "reasons" why...

1) Just didn't get to it.

2) The routines are not thread-safe, yet existing usage by things like
    Tclkit demand their publication in stubs tables.  (Has problems
    similar to those of Tcl_SetMainLoop() (see 801429))  Would be best to
    clean these matters up first.

3) Wanted to combine with a larger "reform Tcl_Main()" effort.

4) Even though I've beat the drum as hard as anyone to decouple Tk, I
    shy away a bit from actually doing it, since it removes a useful
    testing arena.  I'm shy about closing off the ability to let Tk
    have access to routines like TclParseNumber() or
    TclBackgroundException() where doing so might help prove those
    interfaces before we make them public.

5) Even with no more tclInt.h use, Tk will still be some distance away
    from being "just another package" due to lingering legacies of its
    history of assuming "I'm embedded in (a big) wish".

All that said, I support your goal.  Can you generate a patch and
attach to Tk Feature Request 1336531 where we can hash out the
remaining details?

-- 
| Don Porter          Mathematical and Computational Sciences Division |
| don...@ni...             Information Technology Laboratory |
| http://math.nist.gov/~DPorter/                                  NIST |
|______________________________________________________________________|