Re: [TCLCORE] TIP #257: Object Orientation for Tcl

[Lars <lar...@re...>]

At 10.17 +0200 2005-09-27, Donal K. Fellows wrote:
> TIP #257: OBJECT ORIENTATION FOR TCL
>=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> Version:      $Revision: 1.2 $
> Author:       Donal K. Fellows <donal.k.fellows_at_manchester.ac.uk>
>               Will Duquette <will_at_wjduquette.com>
>               Steve Landers <steve_at_digitalsmarties.com>
>               Jeff Hobbs <jeffh_at_activestate.com>
>               Kevin Kenny <kennykb_at_users.sourceforge.net>
>               Miguel Sofer <mig_at_utdt.edu>
> State:        Draft
> Type:         Project
> Tcl-Version:  8.5
> Vote:         Pending
> Created:      Monday, 26 September 2005
> URL:          http://purl.org/tcl/tip/257.html
> WebEdit:      http://purl.org/tcl/tip/edit/257
> Post-History:
> Obsoletes:    TIP #50

I'll be naughty and raise the traditional "TIP killer" question:

   Is there any reason this /can't/ be done as a package instead?

I'm not particularly worried about the "core download bloat" issue, because
I suspect that the actual amount of code it takes to implement this is
anyway dwarfed by the amount of code for the recent numerics changes (which
I strongly support). Instead there on one hand there is the principle, and
on the other there is the matter of 8.5 feature completeness.

If I recall the release planning for 8.5 correctly the first beta is
getting close, which by the "feature-complete" distinction between alpha
and beta means that an implementation of this TIP would have to be put in
place in time for this, or miss 8.5 completely. For that to happen, the TIP
first needs to get fairly stabilised in what mechanisms it proposes, and
from the look of the present discussion the TIP still has some way to go in
that area. Rushing it in could potentially lead to unpleasant d'ohs next
year, and delaying 8.5 release is perhaps not altogether popular.

If it was a package however, it would be less obtrusive to take a middle
road and let the exact details mature an extra couple of months. There
could be a "core" TIP stating that there will be an OO support package
called "oo" always available and laying down its basic principles, but
giving the exact details only as a suggestion to be carved in stone by a
second TIP. Then if it turns out the preliminary specification will have to
be changed incompatibibly this can be handled gracefully by incrementing
the major version number of the oo package.

But maybe I'm being overly rigid here.


Another issue: Will this mean that
   TIP #180: Add a Megawidget Support Core Package
can be taken out of the freezer? As I recall the discussion of that
proposal, the main argument against was that it would duplicate features in
Itcl which by TIP#50 were already going to be available "in the core", but
since the OO of TIP#257 does not address the configuration issue that
argument now loses its weight.

Lars Hellstr=F6m