Re: [Gnash-commit] gnash/libgeometry Range2d.h

[strk]

On Wed, Feb 28, 2007 at 01:44:47PM +0000, Martin Guy wrote:

> I mean, I still haven't figured which of the three or four different
> interfaces to AS methods is this week's preferred one, or which ones
> work and which don't.

Last is preferred. All should "build", but last is the most robust.

> Every time I see a commit it's using a new improved version different
> from before.

Sorry, this is because things are found in progress. 
I just found out that methods *need* to be instances of Function.
We should add tests for this in the actionscript.all testcases.
For example, not only:

        check_equals(typeOf(String.prototype.slice), 'function');

But also:

        check(String.prototype.slice instanceOf Function);

Or

        check_equals(typeOf(myString.slice.apply), 'function');

> Are you in a position yet to decide on one scheme, apply it everywhere
> some boring afternoon and remove the support code (types etc) for all
> the old ways?

"Maybe" ?
Best way to get to that position is having contribution to testcases trying 
every
evilish code trying to break Gnash (make it abort). Also, good code which is 
expected
to work of course.. but it looks like all people care about is playing 
youtube...

> I mean it's ok to work like if you're the only one working on a
> program, but it's hard for anyone else to hit a moving and seemingly
> confused target.

All existing code should still work, just NOT be robust as the new one.
If there's anything you don't get just ask, or if you prefer I can send a mail 
*before* 
committing such changes for a prior discussion.

Note that I introduced the the builtin_function class long ago (July 2006) and
I think I explained the reason for that in a mail on gnash-dev. Just didn't 
port all
the classes  to new layout. I was hoping newbies would take that task on, so 
I'm 
very happy if you want to take that over.

--strk;