Robin Leroy

@markusicu, I am running into the same problem as in https://github.com/unicode-org/icu/pull/3028#issuecomment-2233615599: the documentation only tells me how to regenerate the old icudata.jar, not the ICU4J .brk files. Can you perform...

@markusicu post-UTW poke

@markusicu in fcd04fc67b63d23e9be34803d734df18ceb44868 I tried copying over the .brk files that get generated when I rebuild on my machine; these don’t seem to work: ``` java.lang.AssertionError at com.ibm.icu.impl.ICUBinary.readHeader(ICUBinary.java:574) at com.ibm.icu.impl.RBBIDataWrapper.get(RBBIDataWrapper.java:295)...

Thanks a lot. I squashed it all, dropping ca87360595265dc60c9fa11917cbfdb1872e76a7 (but keeping 081efefddb54f072e21f1ecf85fe1c15cd2aa464) to see if the brk files I generated were fine after all—it looks like it works, so I...

@markusicu As discussed over virtual tea, you might still want to flip the bytes.

> please squash again [Squished.](https://youtu.be/OAGfx7eB3gA?t=152)

You mention that the issue occurs when (emphasis mine) > using the `+=` operator on a _subclass of `UnicodeString`_ The documentation explicitly says https://github.com/unicode-org/icu/blob/9f1c57412afbedb180941a3739c5a460beff24f4/icu4c/source/common/unicode/unistr.h#L232 So I think we would want...

(Incidentally, since this has been possible since C++11, we might want to make UnicodeString final, instead of saying that you should not subclass it in the middle of a hundred-line...

This seems like a reasonable use case, and something to consider as part of the broader question of attitude-keeping (#2556); the attitude control law should not change as you look...