Am Dienstag, 6. Mai 2008 06:02:35 schrieben Sie:
>  Thanks, Kevin.
> 
>  On May 5, 2008, at 7:36 PM, Kevin Kenny wrote:
>  > Will Duquette wrote:
>  >> I've become quite fond of it; my view is that the variables *are*
>  >> declared, in the type definition.  In any event, it's now standard
>  >> practice for Snit code, which means that a TclOO-based Snit needs to
>  >> support it.
>  >
>  > Indeed.  And my impression is that we were discussing *how* best
>  > to accomplish this.
> 
>  I'm sorry, did I sound strident?  Didn't mean to.  Trevor had
>  pointed out that implicitly-declared variables weren't particular
>  Tcl-ish, which is true, though he understood that others disagree;
>  and I was explaining that while I do disagree, that's almost
>  irrelevant at this point, given the context.  But I didn't mean
>  to sound argumentative.
> 
>  > Itcl has the same requirement (that instance variables become
>  > visible automatically within method bodies), and Arnulf has been
>  > able to code up support (resorting to C, I concede) to provide
>  > the implicit variable access via custom variable resolution.
>  > I suspect that Snit's requirement is almost exactly the same
>  > as Itcl's, even though Snit's Not Incr Tcl.
> 
>  This is a great observation.  Here I've been, pondering how to
>  make this work appropriately for Snit in the context of inheritance,
>  and of course it's a problem Itcl has already had to solve.  D'oh!
> 
>  Arnulf, how does Itcl handle this now?  (From the point of view of
>  a programmer using Itcl, not in terms of how it's implemented
>  internally.)

Itcl has a set of variables for every object (all variables of all classes of 
that object) and uses a variable resolver to get the object specific set of 
variables (C-Code !!). Also variables with the same name in different classes 
are handled as different variables, depending on the namespace (== class) the 
method using it is runninng. Every method gets the classes namespace where 
the method is defined as the current namespace when calling it.

I hope that helps, if there are further questions please let me know.
Arnulf
> 
>  > I'd be perfectly
>  > willing, if you and Arnulf can sort out the requirement, to
>  > entertain a TIP promoting the code needed to auto-resolve
>  > instance variables to the Core - if two major OO extensions
>  > need almost identical functionality, that argues in favor of
>  > putting it in.
> 
>  Thanks very much!
> 
>  > In the meantime, inserting code at the start of each method
>  > to bring in the instance variables sounds right.  The
>  > [my variable] command has to be made fast in any case, so
>  > the performance shouldn't go horribly astray.
> 
>  This is more or less what Snit 1.x does, although it uses
>  multiple upvar's, and that's been fast enough for our production
>  code.  Slower than Snit 2.x, which uses "namespace upvar", but
>  fast enough to be usable.
> 
>  Will
> 
>  > --
>  > 73 de ke9tv/2, Kevin
> 
>  ------------------------------------------------------------------
>          will -at- wjduquette.com      | Catch our weblog,
>  http://foothills.wjduquette.com/blog | The View from the Foothills



-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Tcl-Core mailing list
Tcl-Core@[...].net
https://lists.sourceforge.net/lists/listinfo/tcl-core