Approved, on behalf of Lorien. Reviewed by Jim over my shoulder.
Max Carlson wrote:
Change 20070607-maxcarlson-B by [EMAIL PROTECTED] on 2007-06-07
15:15:46 PDT
in /Users/maxcarlson/openlaszlo/trunk
for http://svn.openlaszlo.org/openlaszlo/trunk
Summary: Allow history system to initialize history wrapper.
New Features:
Bugs Fixed: LPP-4108, LPP-4098
Technical Reviewer: mcarlson
QA Reviewer: (pending)
Doc Reviewer: (pending)
Documentation: The history mechanism that traps backbutton events and
allows javascript to set attributes on the canvas, relies on the h.html
page loading in a particular order /wrt the laszlo application, but this
order is not enforced. The result is that when the h.html file loads too
soon, the initial load doesn't register with the app, and when h.html
loads too late the application may attempt to read parameters using
lzSetCanvasAttribute or push history state before h.html is available.
Both cases can cause unexpected behavior.
The fix is to enforce the order in which h.html is loaded in the browser
and the application attempts to interact with the history mechanism. The
changes are:
- Load h.html from LzHistory.as to ensure that the application is ready
to recieve messages from it before it is loaded.
- When h.html has finished loading for the first time LzHistory
generates an onReady event, and sets it isReady attribute to true, to
let the application know that it is ok to push history states and load
attributes from Javascript
The files that are affected in this change are:
- embed.js: lzHistEmbed() no longer writes the h.html iframe to the
page, but instead writes a <div> that the h.html iframe can be written
to later
- embed.js: contains a new method called lzWriteHistoryIframe that is
called by LzHistory.as to load h.html into an iframe in the <div>
created in lzHistEmbed()
- LzHistory.as: calls lzWriteHistoryIframe() when it loads
- LzHistory.as: Generates a onReady event the first time h.html loads,
has an isReady flag that indicates if it is in a ready state
Release Notes: - The interfaces in embed.js remain the same, so the
wrapper html doesn't have to change.
- Applications that make use of the history mechanism will need to wait
for the onReady event and/or check the isReady state before pushing
history states
- Applications that read attributes using lzSetCanvasAttribute need to
wait for the onReady event and/or check the isReady before reading the
attributes
- The onReady event may be generated before the application attaches a
listener to it, so you should not rely on this event firing to indicate
readiness. You should check the isReady flag, and if it returns false,
then attach the event listener to the onReady event
Details:
Tests:
Files:
M WEB-INF/lps/lfc/services/LzHistory.as
M lps/includes/embed.js
Changeset:
http://svn.openlaszlo.org/openlaszlo/patches/20070607-maxcarlson-B.tar
--
Regards,
Max Carlson
OpenLaszlo.org
