I noticed an implementation difference between ICU4C and ICU4J. ICU4C accepts ISCII, and doesn't allow ISCII as an alias. ICU4J alias resolution works differently, and ISCII is not a supported converter alias.
When I enable these tests in ICU4C and port the changes over to ICU4J, some tests fail.
Index: test\testdata/conversion.txt
===================================================================
--- test\testdata/conversion.txt (revision 22412)
+++ test\testdata/conversion.txt (working copy)
@@ -444,7 +444,7 @@
{ "UTF-8", :bin{ 61F48FBFBF }, "a\U0010FFFF", :intvector{ 0, 1, 1 }, :i
nt{1}, :int{0}, "", "", :bin{""} }
// Headers { "charset", "bytes", "unicode", "offsets", "flush", "f
allbacks", "errorCode", "callback", "invalidChars" }
{
- "iscii",
+ "iscii-dev",
:bin{ EF4BC0E9BFE9E8D80AEF4AC0D4BFD4E8D80AEF4838B30AEF4939B30AEF4A3
AB30AEF4B3BB30A},
"\u0A22\u0A3C\u0A5C\u0A4D\u0A39\u000A" /* Gurmukhi test */
"\u0AA2\u0AB5\u0AA1\u0AB5\u0ACD\u0AB9\u000A" /* Gujarati te
st */
@@ -1084,7 +1084,7 @@
// escape callback
{
- "ISCII",
+ "iscii-dev",
"A\u0901\U00023456\u0902B\U00023456C",
:bin{ 41ef42a1255544383444255544433536a24225554438344425554443353643
},
:intvector{
