Maybe `trim` should not be idempotent

[conjugateprior]

Applying trim to a dfm should probably have different minDoc and minCount defaults. In particular, not minCount=1. Combined with a lack of info message, this gives the user the impression that something has happened when nothing actually has.

[kbenoit]

Well, the info message is on by default. And the question is, what should be the defaults? Should there even be defaults?

[conjugateprior]

No info message in 0.9.7-17:

> gg <- dfm(inaugCorpus)
Creating a dfm from a corpus ...
   ... lowercasing
   ... tokenizing
   ... indexing documents: 57 documents
   ... indexing features: 9,214 feature types
   ... created a 57 x 9215 sparse dfm
   ... complete. 
Elapsed time: 0.221 seconds.
> hh <- trim(gg) # *crickets*
> 

[adamobeng]

@conjugateprior is right, the "Removing features occurring fewer than" message only triggers if minCount > 1.