RE: Class Objects Usage

[Fred Schuld]

I realized after I had sent the email that the function in question was defined as a virtual function in ov-fcn.h oops!
(This is my first look at the octave code).

Anyways, I shall try a rebuild with the call stack code fix and check it out.

BTW, is there support for inheritance with class objects?
The last I saw mention of class objects in the mailing list was last year during the summer of 2007 when it was introduced - at that time (I think) inheritance was not supported yet.

Regards,

Fred Schuld

> Date: Tue, 8 Jul 2008 15:28:47 -0400
> From: address@hidden
> Subject: Re: Class Objects Usage
> To: address@hidden
> CC: address@hidden; address@hidden; address@hidden
>
> On 8-Jul-2008, Michael Goffioul wrote:
>
> | On Tue, Jul 8, 2008 at 9:10 PM, Michael Goffioul
> | <address@hidden> wrote:
> | > I can reproduce the problem. I'll have a look.
> |
> | I debugged a little bit with my code, which corresponds to the graphics
> | archive (not the latest octave archive, but quite close). The problem is
> | that in Fclass, fcn->is_class_constructor() returns false (I use the same
> | polynom example). At that point, fcn is actually the "class" function of
> | type octave_builtin, while I think the expected value is the "polynom"
> | function. I stepped into the previous octave_call_stack::caller() call and
> | it appears that the callstack contains 2 elements:
> | [0]: polynom
> | [1]: class
> | and octave_call_stack::caller() actually returns the second one.
> |
> | Might this be related to the uncommited change in octave_call_stack
> | that you mentioned in another mail?
>
> Yes, it could be. I think I checked in at least one intermediate set
> of changes while I was fixing the call stack code. I'd suggest
> updating and see if that fixes the problem. If not, then we can look
> deeper.
>
> jwe
>

---

Use video conversation to talk face-to-face with Windows Live Messenger. Get started.