Behaviour of normalizeTagNames settings for French language locators

Our framework supports English and French. We are using French locators (based upon text, id etc.) to locate the elements in Android.

For last two nightly runs we are having ill-formed XML exception for Android. On our local machines we are not able to reproduce the issue. For our nightly machines, we need to put a fix for this.

If we put the fix mentioned in above thread then there is a possibility that French test cases will start failing since JUnicode will generate the xml for French without any accent characters and locators will fail and hence test will fail.

Can anyone please guide me how can I fix it in a way that should work independent of any language