I hope I've posted this in the correct forum. After looking at other posts it seems like it would get moved here even if I posted in the dijit forums initially.
I have a significant amount of experience with Dojo and am currently developing a set of widgets that will allow users to configure shipping options for eBay listings with a third party listing application. I will see if I can post a link to the application if needed, but for right now it easier to just describe the problem.
In fact, realizing how much I would have to write and not wanting people to stop reading because there is too much information, I'll provide a high level description of the code and the problem and provide more information as needed. I include CustomWidgetA in a large web page (which also contains an instance of Spaw RTE), additionally, I embed several ItemReadDataStores in the HTML. CustomWidgetA has a few form "dijits" and one instance each of CustomWidgetB and CustomWidgetC. CustomWidgetB and CustomWidgetC have a few form "dijits" and 3 instances of CustomWidgetD and CustomWidgetE, respectively. CustomWidgetD and CustomWidgetE also contain several form "dijits". Some of the form "dijits" I use are FilteringSelects that interact with the ItemReadDataStores to retrieve dynamic options from the server.
So, to describe the problem, everything works fine in FF in both the actual page and in a test page where the only content is my widgets. However, I am getting behavior that I cannot explain or debug in IE7. I know the problem is similar in IE6, but the information below is only verified in IE7. I have the following information:
1. In the actual page nothing is rendered in the area where my widgets should appear.
2. In the test page, CustomWidgetA is partially rendered, displaying its form "dijits" and CustomWidgetB however CustomWidgetC is not rendered at all and I get a JS error saying that the variable in CustomWidgetA that refers to CustomWidgetC is "null or not an object".
3. If I remove CustomWidgetA from the test page and replace it with CustomWidgetC, CustomWidgetC partially renders and only shows one instance of CustomWidgetE with no JS error.
4. If I do the same thing as in #3 but remove "isDebug: true" from djConfig, all three instances of CustomWidgetE appear. But removing this from djConfig does not change the results of #1 or #2.
I have googled extensively trying to find a way to fix this and have found nothing. I did find several threads about the 'extra comma' issue in IE and am confident that this is not my problem. My best hypothesis is that this problem is caused by some asynch issue in IE related the the dojo parser and the onload event. However, I have no idea how to proceed from here.
Any assistance will be greatly appreciated.
Thanks,
Mike Daniels

Hard to tell from the info
Hard to tell from the info presented, but if you are using dojo functions before the page and dojo have fully loaded, you would probably see similar symptoms to what you describe--could be timing sensitive. To avoid this, have those functions be called from the function referenced in dojo.addOnLoad(func ref);, except for those functions for which you are 100% positive can only be accessed after all dojo loading and widget creation...
Also, if you are loading script into dijit.layout.ContentPanes via setContent or setHref, that script may not run. An alternative is dojox.layout.ContentPane.
Are you calling a startup()/render()/layout() method on your custom widgets to force them to display?
When one or more widgets do not appear, do they appear if you resize the browser?
Problem fixed - solution may be interesting to someone
So I actually had two problems here. The less interesting one was a syntax error in one of my widget HTML template files that FF was not catching but IE was. In my experience that is really bizarre because in the past I always found IE to be more forgiving of HTML syntax issues.
The other problem is actually more interesting and may be of interest the dojo community. This problem could not be recreated in my test page. It could only be seen in the full HTML page where other JS libraries were being loaded so my guess that external javascript could be required to recreate this issue. The problem was that dojo.require statements would not successfully load any widgets. I fixed it by applying the fix described in this TRAC bug: http://trac.dojotoolkit.org/ticket/6096. I noticed that the bug was closed and I don't know if this change will be caught in a future release, but this fix is definitely required for my application.
Hope this helps someone else!