Actually, I take it back. Only the pseudo-module names are determined in the service layer - instantiation occurs in an observer. Still, tI'm concerned about the messed-up parentApplication.
On Tue, Jul 29, 2008 at 11:33 AM, Richard Rodseth <[EMAIL PROTECTED]> wrote: > Good suggestion. For good measure (groan) I added a few overrides of > measure() that call super, that I could set breakpoints in. > Recall that the hierarchy is like this: > > TiledCanvas > PodChrome > ReportModuleA (not a real module, just a VBox) > ReportChrome > ReportA > Grid > > I notice that ReportModuleA.measure() does not get called. I also > noticed in the debugger that ReportModuleA.parentApplication is not > set (the debugger gives a 1009 error) when you select that property). > Seems suspect, doesn't it? > > ReportModuleA and PodChrome are instantiated in ActionScript, and > ReportModuleA is actually instantiated in the service layer via a > static method in its MXML file. Then a parent of TiledCanvas observes > a list of the pseudo-modules, wraps them in chrome and adds them to > the TiledCanvas. Seemed reasonable at the time. > > On Mon, Jul 28, 2008 at 9:46 PM, Alex Harui <[EMAIL PROTECTED]> wrote: >> You'll have to walk through the measure() methods on each container to see >> what it is computing and why. >> >> >> >> ________________________________ >> >> From: [email protected] [mailto:[EMAIL PROTECTED] On >> Behalf Of Richard Rodseth >> Sent: Monday, July 28, 2008 2:19 PM >> To: [EMAIL PROTECTED]; [email protected] >> Subject: [flexcoders] Re: Measurement and template component >> >> >> >> Oops. Meant for flexcoders. >> >> On Mon, Jul 28, 2008 at 2:18 PM, Richard Rodseth <[EMAIL PROTECTED]> wrote: >>> Turns out to be unrelated to the fact that ReportChrome is a template >>> component (the same thing happens if it's a plain Canvas). >>> Removing width="100%" height="100%" on ReportChrome results in the >>> TiledCanvas tiling correctly but then I am forced to set >>> minWidth/Height on ReportA or its children, or they shrink to nothing. >>> >>> I'm not sure I understand this - I would expect that the measured >>> values would bubble up regardless, and that the measured width/height >>> of a Canvas or subclass thereof (ReportChrome in this case) would be >>> the maximum of the measured children or the specified container >>> percentage. >>> >>> On Wed, Jul 23, 2008 at 5:18 PM, Richard Rodseth <[EMAIL PROTECTED]> >>> wrote: >>>> My application has a structure like this, portions of which are >>>> created dynamically: >>>> >>>> TiledCanvas >>>> PodChrome(1) >>>> ReportModuleA >>>> ReportChrome(1) >>>> ReportA >>>> Grid >>>> PodChrome(2) >>>> ReportModuleB >>>> ReportChrome(2) >>>> ReportB >>>> Chart >>>> etc. >>>> >>>> PodChrome and ReportChrome are template components. >>>> TiledCanvas is functional - when tiling it correctly reflects a >>>> minWidth/minHeight set way deep (eg. on the Grid). But only if I omit >>>> the ReportChrome wrapper. ReportChrome is just a Canvas with two >>>> states, that does an addChild in init(). I added a call to >>>> this.invalidateSize() and child.invalidateSize() after the addChild() >>>> to no avail. Any ideas what I could be doing wrong? >>>> >>> >> >> >
