William,
There is most definitely something more to widgets than just the XML and the attached javascripts. Some kind of "server-side" (or "robohelp-side" I guess) processing for at least some of these widgets.
Take, for instance, the "topic" widget. Its contents does not explain how RoboHelp takes data from the original source XML and displays it. Here's the contents:
<widget displayname="IDS_TOPIC" category="topic" popupmenu="IDS_TOPIC" ><copyfiles> <copy srcdir="../common/scripts" destdir="template/scripts"> <file src="mhtopic.js" dest="mhtopic.js" link="true" patch="true" unique="false"/> </copy></copyfiles><script><![CDATA[
gTopicId = "{{MSM_PAGE_DATA.TOPIC_ID}}";
]]></script><placeholdercode><![CDATA[ <?rh-msp-topic class="wTopic" widgettype="topic" ?>
]]></placeholdercode></widget>
So basically we have 1 JS file being copied, one JavaScript using a super-secret undocumented variable called "MSM_PAGE_DATA.TOPICID" (which returns ZERO information on google) and the placeholder expected for the widget.
There is a lot of openness and information sharing missing between Adobe and their clients. "Undocumented" features and tools will be the end of me, I swear.