As I said, I didn’t know how the ‘interaction’ happened, only that it did.
on the other hand (if this isn’t halted by feasibility issues) OpenOffice’s licensing system should allow copying/reuse of the components of the spellchecker that are responsible for accessing/updating the user dictionaries (modified to LT’s internal needs).
Good news: without hard-coding the file path of my file. It works.
So this is a pure Java problem. Something like: finding files that are outside the package.
BTW, I saw in UserConfig.java there is a userSpecificSpellerWords which is added to the ‘ignore list’ here. Modifying this part seems to me cleaner.
Finally, I replaced the ignore.txt by a symbolic link to my vim file.
Then I pasted the original lines in ignore.txt into this file (in case some of them are required in LT ?).