From: Donal K. Fellows <fellowsd@cs...> - 2000-10-27 23:51:11
John Ousterhout <ouster@...>
> After I sent my message last night I had the same thought myself. If
> we bundle [incr Tcl], it should just become part of the core: there
> should be no need to say "package require ..." to get the "class"
> command. If you think about the arguments in favor of bundling ("Tcl
> isn't serious about OO", etc.) keeping the appearance that [incr Tcl]
> is a separate package will still leave us vulnerable to these criticisms
> ("they ship it with the core, but you can't use it unless you do extra
> work; no, Tcl *still* isn't serious about OO").
I've been thinking a bit about this (in between sorting out the TIPs
and attending seminars) and I reckon that itcl is pretty good as far
as OO packages go (going by what I read in my copy of Tcl/Tk Tools.)
It has a clear and obvious syntax for declaring classes, and adding an
OO organizational pattern to the language would certainly benefit us
in terms of mindshare. My concerns are:
Should we crowd out other OO systems? *Any* kind of core blessing
for itcl will do this (by preponderance of specific code in
libraries, if nothing else) though I suppose we're damned if we do,
and damned if we don't. This is still my main concern.
Should our objects support access control? It would be an anomaly
if we do.
Should we permit naming of objects? Should we *force* it (the
current situation if I'm not mistaken?) This has proved to be a
problem in the past with Tk's images, though the problem is greatly
reduced if you can't use a command name that is currently in use.
Donal K. Fellows, Department of Computer Science, University of Manchester, UK.
(work) fellowsd@... Tel: +44-161-275-6137 (preferred email addr.)
(home) donal@... Tel: +44-1274-401017 Mobile: +44-7957-298955
http://www.cs.man.ac.uk/~fellowsd/ (Don't quote my .sig; I've seen it before!)
The TclCore mailing list is sponsored by Ajuba Solutions
To unsubscribe: email tclcore-request@... with the
word UNSUBSCRIBE as the subject.