Fixing incorrect detection of the NewLisp language

[Ignotus]

The proposed sample file that contains tokenized Dutch words has been wrongly recognized as a NewLisp code in my repository.

This patch is a try to fix this ambiguity.

[pchaigno]

Isn't the classifier sufficient to solve this issue? I.e., does adding new samples solve it?
Otherwise, have you tested that this new heuristic doesn't break recognition of NewLisp files?

[Ignotus]

Yeah, you can try to check

bundle exec linguist samples/Text/corpus_1000.nl

No, I only tested on unit tests.

[pchaigno]

Yeah, you can try to check

bundle exec linguist samples/Text/corpus_1000.nl

The files used for training should always be recognized correctly (otherwise there's something really wrong with the classifier). Do you have some other files wrongly detected as NewLisp?

[Ignotus]

No, unfortunately. I have fixed in my project by adding .gitattributes. Probably I will create another pull request in the case if I will have enough data to test in the future.