Second, I am seeing short radicals on nested situations with FF3.6.12
with STIX 1.0 but did not find this in the bug list. You can see an
example with the torture test on mrootB1 (note that the radical is too
short) and here https://eyeasme.com/Joe/MathML/MathML_browser_test.html
(test 2) - you can see that with FF 3.6.12 the radicals in test 2 are
too short and are not consistent with the example for FF3.6 with beta
STIX. It has been reported to me that this situation is not present
in beta 4
STIX fonts 1.0 are not supported by Firefox 3.6, you need to use the
STIX beta fonts:
https://developer.mozilla.org/en/Mozilla_MathML_Project/Fonts#STIX_Fonts
Regarding Firefox 4, there is a workaround to stretch operators without
STIX fonts installed (this works well for radicals)
https://bugzilla.mozilla.org/show_bug.cgi?id=414277
and STIX fonts 1.0 are supported:
https://bugzilla.mozilla.org/show_bug.cgi?id=569195
(it has been proposed to take the latter bug for Firefox 3.6 but this is
waiting approval by the drivers).