Before I can review this, I _really_ have to understand what the contract of getAttributeRelative is. The documentation appears to contradict itself:
> * Transform the specified attribute (width/height/x/y) from the existing > * coordinate space to the reference view's coordinate space. This > * transformation takes into account any nested views between the views and > * the common ancestor view. > * > * This method answers the question: what should this view set its > * width/height/x/y to in order to appear to have the same value for that > * attribute as the reference view? The first paragraph would appear to say, e.g., for `x`, "if I move to x=0 in this view, what value of x in the reference view will get to the same absolute position". The second paragraph seems to specify the same thing with this view and the reference view reversed. Perhaps we can intuit what people expect it to do and correct the documentation? Kathryn's test case implies to me the contract is: For x/y: What value of x/y in the reference view will be the same absolute position as my current x/y? For width/height: What distance in the x/y axis in the reference view will be the same absolute distance as my current width/height. Is that a contract we can all agree on? On 2010-09-02, at 17:59, Max Carlson wrote: > Change maxcarlson-20100902-vX9 by maxcarl...@friendly on 2010-09-02 14:52:06 > PDT > in /Users/maxcarlson/openlaszlo/trunk2 > for http://svn.openlaszlo.org/openlaszlo/trunk > > Summary: Improve getAttributeRelative() > > Bugs Fixed: LPP-9333 - lz.view getAttributeRelative() doesn't work with > nested views inside a view that has xscale/yscale > > Technical Reviewer: ptw > QA Reviewer: [email protected] > > Details: lztest-lzview.zlx - Add tests for getAttributeRelative(), clean up > older tests to be a bit more tolerant. > > LzViewLinkage - downlink and uplink arrays now include the actual node, so > scaling computation is simpler. Use correct algoritm for determining scale > and offsets. Add typing. > > LaszloView - Test for and return value immediately when refView === the > curent view. LzViewLinkage now includes the current view in the computation, > so there's no need to factor in its value here. > > Tests: lztest-view.lzx runs as before in all runtimes - with fewer warnings. > This highlights issues with resourcewidth in swf10, but that's another ball > of wax. > > Files: > M test/lztest/lztest-view.lzx > M WEB-INF/lps/lfc/views/LzViewLinkage.lzs > M WEB-INF/lps/lfc/views/LaszloView.lzs > > Changeset: > http://svn.openlaszlo.org/openlaszlo/patches/maxcarlson-20100902-vX9.tar
