Yes it was an issue - please verify it with 3.3.3 Beta1. Should be already fixed.
I did some more testing with 3.3.1.GA and 3.3.3.BETA1, the result is the same.
I did some debugging too on 3.3.2.SR1, compared the calendar's decode process with a simple h:inputText and it seems that there is no disabled flag check in the CalendarRenderer's decode function (it uses the inherited decode from RendererBase). In the h:inputText's renderer (HtmlBasicRenderer) there is a shouldDecode function and call.
Tried to find a jira entry about this bug without success. Is there a fix available somewhere? I need it backported into 3.3.2.SR1.
Thanks for investigation, i checked and reopened https://jira.jboss.org/jira/browse/RF-7746 . I highly recommend you to use 3.3.3 release after fix because 3.3.2 SR1 will not be patched with this fix and has also one additional regression issue with menus (already fixed in 3.3.3). In general if you stucked at some RF version - unfortunatelly you will have to patch the release jars on your own because we can't support so much branches at the time. You could check enterprise support option if need such issues fixes/backports.
I think switching to 3.3.3 won't be a problem especially if the GA version is released in a few months.
Thank you for your cooperation