No registered users in community xowiki
in last 10 minutes
in last 10 minutes
[Xotcl] Re: xotcl and itcl ...
From: Gustaf Neumann <neumann_at_wu-wien.ac.at>
Date: Sat, 27 Mar 2004 16:24:42 +0100
On Tuesday 23 March 2004 23:46, Jeff Hobbs wrote:
> I was intrigued by the recent posting about an itcl emulation
> using xotcl. How much work would it be to get a 100% emulation
> of itcl in xotcl? After all, there is tcl++, that was a pure
> Tcl emulation of itcl (that was 99.9% IIRC).
Jeff,
This is certainly do-able, especially for well structured
programs, but should be done by a person with
a reasonable itcl knowledge and a real-world test base.
We can certainly help in such a project.
in genereal i see two kinds of issues to be adressed
- some ideosyncaries of itcl, which can be easy
emulated by procs (e.g. the auto nameing convention
of itcl), but this would slow down the code.
- the apporach of http://mini.net/tcl/10975
works fine for well-structured itcl programs
(calls along the class chain are in the right
order, calls do not jump out of the class chain,
etc.); when these should be supported, we
have to look for a good approach.
Btw., the code in http://mini.net/tcl/10975
can be further improved by replacing the
redirector procs by interp aliases. this
speeds up the objinst benchmark by
further 4-5 percents...
-gustaf
Date: Sat, 27 Mar 2004 16:24:42 +0100
On Tuesday 23 March 2004 23:46, Jeff Hobbs wrote:
> I was intrigued by the recent posting about an itcl emulation
> using xotcl. How much work would it be to get a 100% emulation
> of itcl in xotcl? After all, there is tcl++, that was a pure
> Tcl emulation of itcl (that was 99.9% IIRC).
Jeff,
This is certainly do-able, especially for well structured
programs, but should be done by a person with
a reasonable itcl knowledge and a real-world test base.
We can certainly help in such a project.
in genereal i see two kinds of issues to be adressed
- some ideosyncaries of itcl, which can be easy
emulated by procs (e.g. the auto nameing convention
of itcl), but this would slow down the code.
- the apporach of http://mini.net/tcl/10975
works fine for well-structured itcl programs
(calls along the class chain are in the right
order, calls do not jump out of the class chain,
etc.); when these should be supported, we
have to look for a good approach.
Btw., the code in http://mini.net/tcl/10975
can be further improved by replacing the
redirector procs by interp aliases. this
speeds up the objinst benchmark by
further 4-5 percents...
-gustaf
-- Univ.Prof. Dr.Gustaf Neumann Abteilung für Wirtschaftsinformatik WU-Wien, Augasse 2-6, 1090 Wien