only once #838, #837, and #836 are solved, I propose we change the main matching arguments to pattern, from features or keywords.
So this would be:
Also:
Why make this change?
- more consistent across functions
- more descriptive than features, which is really specific to a dfm construct, but we have used it to refer to tokens, sequences, and pattern matches generally
pattern is consistent with the usage stringr and stringi
only once #838, #837, and #836 are solved, I propose we change the main matching arguments to
pattern, fromfeaturesorkeywords.So this would be:
tokens_select(x, pattern)tokens_remove(x, pattern)tokens_compound(x, pattern)dfm_select(x, pattern)dfm_remove(x, pattern)fcm_select(x, pattern)fcm_remove(x, pattern)kwic(x, pattern)Also:
features(andkeywordsfor kwic) arguments in the same way that we did this for the camelcase arguments intokens(). So they will still work but issue a message.pattern, and update the existing one forphrase(and these might be the same page)Why make this change?
patternis consistent with the usage stringr and stringi