Build: 1.5.0_U8 b03
Platform: Windows Vista 5384 32bit


1)Install 32bit version of this JDK build on Wind-Vista 5384 platform
2)Set the JAVA_HOME & PATH variables
3)Set the locale to hindi
4)Use any jdk-demo apps: go to $JAVA_HOME/demo/jfc/Notepad application
5)Input the following combinations of text

[I] Input using the Unicode Hex values directly:
(a) \U0924 + \U094D + \U0930
(b) \U0936 + \U094D + \U0930
(c) \U0915 + \U0948 + \U0938 + \U0947 [should form word called "kaise". Note that the
\U0948 is not getting formed correctly with other chars]

OR

[II} In INSCRIPT mode of Hindi input , you may directly press the following Keys to see the same problems:
(a) Press " l + d + j " : these 3 should form one glyph [The same glyph can be directly entered by just presssing the key " ^ " in Inscript mode]
(b) Press " M + d + j " " these 3 should form one glyph[The same glyph can be directly entered by just presssing the key " * " in Inscript mode]
(c) Press "k + w + m + s" : this should result in "kaise"


You will notice the failure in the formation of these glyphs.
So far the problem was observed mainly when we have combination with [\U094D + \U0930] and [\U0948]

NOTE:
1)You may execute the testsuite " gs15UR-b03.tar " using the testlist for hi_IN locale also & see the problems in output for all the testcases
2) I have attached some snapshot JPG files from these testcases---the image shows the output on left-hand-side with golden image(or expected result)on the right-hand side in this image. The problems are marked in these files.

NOTE:
Tested on Solaris platform & NOT reproducable.