Re: [Gnash-commit] gnash ChangeLog testsuite/misc-ming.all/Makefil...

[zou lunkai]

> mc3 was placed in frame 4 and still alive, we're jumping to frame 5
> so why should we build it again ? I guess its just the 'ratio' thing
> preventing that, right ?

expected behaviour:  mc3 constructed only *ONCE*, never unloaded.
Gnash behaviour: mc3 is unloaded and constructed again.

None of the attempts could solve this, IIRC. 'ratio' here is not
important. Note the ratio of mc1, mc2 and mc3 are different, which is
a common case(sane test).  Updating the ratio of mc3 to any other
value won't change the behaviour of mc3.

--zou


On Nov 23, 2007 3:11 PM, strk <address@hidden> wrote:
> On Fri, Nov 23, 2007 at 02:29:02AM +0000, Zou Lunkai wrote:
>
> > +/*
> > + * Timeline:
> > + *
> > + *   Frame  | 1 | 2 | 3 | 4 | 5 | 6 |
> > + *  --------+---+---+---+---+---+---+
> > + *   Event  |   |P  |RP | RP| * | J |
> > + *
> > + *  P = place (by PlaceObject2)
> > + *  R = remove (by RemoveObject2)
> > + *  J = jump
> > + *  * = jump target
> > + *
> > + * Description:
> > + *
> > + *  frame2: place mc1 at depth 100 with ratio 0
> > + *  frame3: remove mc1, place mc2 at depth 100 with ratio 0.1
> > + *  frame4: remove mc2, place mc3 at depth 100 with ratio 0.2
> > + *  frame5: do nothing
> > + *  frame6: jump to frame5
> > + *
> > + * Observed behaviour:
> > + *
> > + *   (1)mc1 constructed and unloaded twice;
> > + *   (2)mc2 constructed and unloaded twice;
> > + *   (3)mc3 constructed only *ONCE*, never unloaded.
> > + *
> > + * Deduction:
> > + *
> > + *   we need a temporary displaylist when jumping back, otherwise cann't 
> > keep mc3 alive
> > + *   in the current design attempt.
> > + *
> > + */
>
> Doesn't one of the redesign attempts account for this ?
> One was: "keep info about which static character belongs to each frame"
> mc3 was placed in frame 4 and still alive, we're jumping to frame 5
> so why should we build it again ? I guess its just the 'ratio' thing
> preventing that, right ?
>
> --strk;
>