The one thing COERROR is used for is saying that "this word is not the central word of the error". We have e.g. a word1 and word2, an error tag &error on word1 and on word2, and a relation RIGHT from word1 to word2. That should be enough to say that word2 is implicitly COERROR and shouldn't have its own separate underline.
The one thing COERROR is used for is saying that "this word is not the central word of the error". We have e.g. a word1 and word2, an error tag &error on word1 and on word2, and a relation RIGHT from word1 to word2. That should be enough to say that word2 is implicitly COERROR and shouldn't have its own separate underline.
And might make confusions like https://github.com/giellalt/lang-smj/issues/38 less likely.