[Laszlo-dev] For Review: Change 20071214-ptw-i Summary: Make the memory tracer generic

[André Bargull]

Approved!

notes:
- Debug.isMarked has had special "instanceof MovieClip" code, was it 
removable?
- __LzLeaks#initialize(..): missing non-"try..catch" block for $as2
- Debug.traceStep: doc-comment could be updated
- Debug.traceStep: comment "The setInterval is 1400ms (because we know 
scriptRunningSlowly defaults to 1500)" does not match code
- Debug.initTrace: add comment about 1400ms interval-value, like in 
Debug.traceStep
=> What about adding "backgroundTaskInterval=1400" to Debug instead of 
having hard-coded values?

On 12/15/2007 12:51 AM, P T Withington wrote:
> Change 20071214-ptw-i by ptw at dueling-banjos.local on 2007-12-14 
> 18:23:23 EST
>     in /Users/ptw/OpenLaszlo/ringding-2
>     for http://svn.openlaszlo.org/openlaszlo/trunk
>
> Summary: Make the memory tracer generic
>
> New Features: Memory tracing now works (for some definition of works)
> in DHTML in addition to SWF.
>
> Technical Reviewer: a.bargull at intensis.de (pending)
> QA Reviewer: promanik at laszlosystems.com (pending)
>
> Documentation:
>     Debug.markObjects, findNewObjects, whyAlive now work in SWF and
>     DHTML.  DHTML is not completely usable yet because it appears that
>     some underlying runtimes themselves leak DOM objects (or our usage
>     of them causes them to be leaked).  It works reasonably well in
>     Safari, less well in Firefox, hangs Opera, is untested in IE (but
>     IE has its own leak tools).
>
> Release Notes:
>
> Details:
>     kernel.js: Initialize the debug window early so memory tracing can
>     find it.  Use a static style for debugger output, so that a new
>     CSSStyle object is not consed for every line of debugger output.
>
>     LzMemory.lzs: Make this generic (although there are still some if
>     ($as2) bits for now, I think a unified code base is better than
>     splitting this into the kernels).  Add some metering to see how
>     the background tracer is working.  Add some documentation.  Add
>     some try/catch blocks to ignore errors trying to trace native
>     wrapped objects in DHTML.  Use set/clearInterval instead of the
>     old tracer movieclip to run the background task.  Fix the path
>     evaluator to be correct and platform-neutral.  Use
>     objectOwnProperties to find slots to trace.  Make the code to hide
>     the debugger from the leak detector generic, hide the debugger
>     DOM tree.  Make whyAlive output all relevant data, so it can be
>     invoked as a button, not just from the debugger.
>
>     Library.lzs: switch from the platform leak detector to the generic
>     one.
>
>     LzRuntime: Use static style for debugger output.
>
>     laszlo-debugger.css: Define that static style.
>
> Tests:
>     The following test program works for me (pressing the 3 buttons at
>     the appropriate times, creates a leak report):
>
>     <canvas debug="true">
>       <simplelayout axis="x" />
>       <button onclick="Debug.markObjects()">Mark</button>
>       <button onclick="Debug.findNewObjects()">Find New</button>
>       <button onclick="Debug.whyAlive()">Why Alive</button>
>     </canvas>
>
>
> Files:
> D      WEB-INF/lps/lfc/debugger/platform/swf/LzMemory.as
> M      WEB-INF/lps/lfc/debugger/platform/dhtml/kernel.js
> M      WEB-INF/lps/lfc/debugger/LzMemory.lzs
> M      WEB-INF/lps/lfc/debugger/Library.lzs
> M      WEB-INF/lps/lfc/compiler/LzRuntime.lzs
> M      lps/includes/laszlo-debugger.css
>
> Changeset: 
> http://svn.openlaszlo.org/openlaszlo/patches/20071214-ptw-i.tar
>