Idea: Switch orth / orth_, tag / tag_, etc from int / unicode to unicode / int (please debate) about spacy HOT 10 CLOSED

I vote to move the integers to the underscored attributes. I use them less often, so it will save my left pinky a lot of work!

from spacy.

I agree, I think it would be more intuitive to have the integer value as the underscored name rather than as the main name.

from spacy.

👍 on swapping them.

from spacy.

I don't think they should be swapped. While I would prefer to have ints underscored, changing it now breaks the API, and it is really not a very difficult convention to get used to.

from spacy.

I get that it breaks the API, and this is unfortunate. But I'd rather introduce breaking changes now than live with them forever.

from spacy.

Sure...

Also, after some more consideration, I think that having ints as default values for each tag makes a lot of sense since most of the time it's ints one wants to work with, and not unicode strings.

from spacy.

Yeah, agreed that the keyword_ should represent ints and keyword should be the string value. It's going to cause some pain to change APIs though 🎱

from spacy.

I guess I would vote for the change as well. When I first started using spaCy, I was initially confused why using .lemma/.lower/etc was giving me a number and not a string as expected. I think it makes more sense for the string value to be named "foo", and the integer value to be named "foo_", especially for new users.

from spacy.

Two weeks ago I would have said "yes, please swap" but now I'm thinking "please keep as is". The reason is less that I got used to it but more that my work focus has shifted from "data exploration" to "data processing". For the exploratory work the underscores where a bit of a pain but for the processing (which I believe will be the bigger portion) the current convention makes perfect sense.

from spacy.

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.

from spacy.