Re: general c++ question

[Shai Ayal]

On 8/31/07, John W. Eaton <address@hidden> wrote:
> On 29-Aug-2007, John W. Eaton wrote:
>
> | On 29-Aug-2007, Shai Ayal wrote:
> |
> | | This way you will loose the easiness of having them automatically
> | | generated in the macro. But it is better and more consistent.
> |
> | Many of them do not force any other action.  They are simply of the
> | form
> |
> |   PROP = val;
> |
> | so they can be generated by the macro.  Of course we will need a
> | separate macro that does not generate the set_PROP function.  Let me
> | take a look at doing at least one or two classes this way.
>
> After spending some time on this, I think we need something more than
> the C preprocessor to do the job.  I'm working on a patch that will
> move graphics.h to graphics.h.in and then use an AWK or Perl script to
> generate graphics.h.  The properties declarations in the graphics.h.in
> file will be something like
>
>     BEGIN_PROPERTIES
>       graphics_handle currentfigure S
>       octave_value visible
>     END_PROPERTIES
>
> The properties will be specified with
>
>   TYPE PROPNAME FLAGS
>
> The optional FLAGS field can be some combination of characters that
> tell the script whether to declare the field "mutable", emit a
> definition and/or declaration for the set_PROPNAME and get_PROPNAME
> functions, etc.  The script will only generate the simple ones
> automatically (the ones for which the body of the function is just
> "PROPNAME = val;" or "return PROPNAME;"). The others will have to be
> written by hand.  With this approach, I think we can also
> automatically generate the set and get functions that work with the
> names of the fields.  There are a few more details to work out, but I
> think I can have something checked in by next week.


Looks good to me. Until the patch is in I will assume the interface to
c++ stays like you outlined before in this thread using the
get_PROPNAME methods. Hopefully this won't change

Shai