flex-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From aYo <...@binitie.com>
Subject Re: RE : FDB has old code ?
Date Fri, 24 Jan 2014 14:51:41 GMT
Je te souhaite mieux

On Fri 24 Jan 2014 15:44:44 WAT, webdoublefx wrote:
> Thanks Scott for having tried :-)
> Btw Im sick today, will propably commit on sunday, the time to clean up things
> Envoyé depuis un mobile Samsung
> <div>-------- Message d'origine --------</div><div>De : Scott Talsma
<scott@talsma.tv> </div><div>Date :24/01/2014  14:26  (GMT+00:00) </div><div>A
: dev@flex.apache.org </div><div>Objet : Re: FDB has old code ? </div><div>
> </div>
> I think  SWD is only relevant for ActionScript 1.0 and 2.0.  Flash files
> with debugging turned on would produce a swd file.
> When debugging was toggled on in the Flash Player (and a debugging swf was
> loaded), a call for the swf's swd would be made.
> I just did a small test, creating a debuggable Flash file; when I compile
> it for AS2.0, the swd file is generated; when targeting AS3.0, no swd file
> is created.
> On Thu, Jan 23, 2014 at 12:58 PM, Alex Harui <aharui@adobe.com> wrote:
>> I don't know the code that well, but the swd part always surprised me.
>> Old flash swfs had swd files, but I didn't think Flex SWFs did.
>> On 1/23/14 9:43 AM, "Frédéric THOMAS" <webdoublefx@hotmail.com> wrote:
>>> Hi,
>>> I noticed some old code, unused code path, old coding pratices in FDB and
>>> more important, it is very slow relative to the FB debugging experience
>>> (FDB: 1mn30s FB: 8s on my large company app to stop at the 1st breakpoint
>>> set in my runtime module), after a talk with Alexander and having dug
>>> into the code and done some experiments, I would conclude the
>>> waitForMetaData() function is not needed anymore, this function basically
>>> was waiting until each swf was loading info.isPopulated() up to 80 *
>>> 250ms = ~20s to be sure the metadata of the Function infos have been
>>> processed, from my experiements, there's no needs anymore for such a
>>> wait, somewhere else in the code, I can read, "if we are a avm+ engine
>>> then we don't wait for the swd to load" and "added by mmorearty on 9/5/05
>>> for RSL debugging" in DManager.java when getting the Message
>>> DMessage.InSwfInfo it considers the swf is completely loaded
>>> info.setPopulated();
>>> I don't pretend to understand everything in those classes but my guesses
>>> meet my experiments.
>>> Now, the waiting time in FDB / IntelliJ is the same than in FB, meaning
>>> divided by more than 11 in my case.
>>> Does someone know more about the player mechanism to populate the
>>> Function metadatas ?
>>> I'm about to clean my changes, open a ticket and propose this modified
>>> FDB for testing, any objections ?
>>> Thanks,
>>> Frédéric THOMAS


View raw message