[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Gnash-commit] gnash ChangeLog server/fill_style.cpp server/fi...
From: |
Sandro Santilli |
Subject: |
Re: [Gnash-commit] gnash ChangeLog server/fill_style.cpp server/fi... |
Date: |
Thu, 21 Jun 2007 12:43:01 +0200 |
On Thu, Jun 21, 2007 at 08:19:55AM +0800, zou lunkai wrote:
> Question: It seems that besides AS objects, GC also takse over more
> general resouces, eg. SWF tags. Will this increase the scan time at
> later collecting step?
Yes, scan time will be increased by the bumber of items required to
be marked as reachable.
I'm just following the ref_counted layout, not adding new classes.
Actually, I'd like to remove some classes from the ref_counted
tree, in particular bitmap_info and character_def (and descendant).
After all, a definition is always in memory anyway (in the CharacterDictionary)
so it should be managed by the CharacterDictionary. The only exception is
for dynamically created sprites (IIRC) which gets a definition created at
runtime
and the drawable for drawing API, which is again a definition breaking the
documentation for a character_def (an "immutable" definition of a character
element).
We used a shape_character_def to avoid changing the interfaces to the backend
renderer
but I've always noted we had a problem there, suggesting we'd always render
*instances*
instead of *definitions*.
--strk;