Sealed classes are in general faster, how much this impacts delegates
could be tested (may be difficult to create useful performance test,
though). To reduce possible backward compatibility issues, 4.8 could be
shipped with a dynamic lz.Delegate class, so only 5.0 will contain the
sealed class.
Here are some resources about performance tuning and sealed classes for
ActionScript3:
http://www.onflex.org/download/AS3Perf.pdf
http://www.onflex.org/ACDS/AS3TuningInsideAVM2JIT.pdf
On 5/29/2010 11:44 PM, P T Withington wrote:
If we feel there is a performance benefit to having sealed delegates and that
it is too much of an API change to seal lz.Delegate, we could implement a
sealed version that is used internally in the LFC and make lz.Delegate an
unsealed subclass.
On 2010-05-29, at 12:58, André Bargull wrote:
This makes me wonder whether it's possible to ship 4.8 with a sealed
lz.Delegate class...
Change 20100529-bargull-BK2 by barg...@bargull02 on 2010-05-29 18:48:34
in /home/anba/src/svn/openlaszlo/trunk
for http://svn.openlaszlo.org/openlaszlo/trunk
Summary: Making delegate a sealed class broke combobox selection
Bugs Fixed: LPP-9038 (OL 5.0.x - SWF10 - Components Sample - The Pistachio
combobox can not work well)
Technical Reviewer: max
QA Reviewer: (pending)
Details:
lz.Delegate is no longer a dynamic class, that means it's no longer possible to
attach custom properties to instances of lz.Delegate. Just changed basecombobox
to store the selection start and end point directly in the instance.
Tests:
see bugreport
Files:
M lps/components/base/basecombobox.lzx
Changeset: http://svn.openlaszlo.org/openlaszlo/patches/20100529-bargull-BK2.tar
