It seems to me you should either use the 'public' cache API (setCSS? applyCSS?) or you should get rid of it. To have some pieces of the code reaching around the API and magically updating the cache directly just seems like asking for trouble. But then, we seem to already have pieces of code that ignore the cache altogether because they think they know better? And some magic CSS properties that specifically defeat the cache, for undocumented reasons?
I'm only approving this because you say it works. I honestly can't figure out what this cache code is or is not doing any more. It's too convoluted. On 2010-07-16, at 21:59, Max Carlson wrote: > Change 20100716-maxcarlson-9 by maxcarl...@friendly on 2010-07-16 18:56:32 PDT > in /Users/maxcarlson/openlaszlo/trunk2 > for http://svn.openlaszlo.org/openlaszlo/trunk > > Summary: Fix bug where sprites will sometimes show after being hidden > > Bugs Fixed: LPP-9201 - DHTML: sometimes clicking the close box on the > debugger causes it to disappear, then reappear > > Technical Reviewer: ptw > QA Reviewer: hminsky > > Details: I'm checking this in ahead of review. > > LzSprite - Store display values to __processHiddenParents doesn't trample > them. Comment outstray debug.warn() calls. > > Tests: See LPP-9201 > > Files: > M WEB-INF/lps/lfc/kernel/dhtml/LzSprite.js > > Changeset: > http://svn.openlaszlo.org/openlaszlo/patches/20100716-maxcarlson-9.tar
