2017.06.01

examples.

2017.01.15

Informative errors when type.list is unusual.

2015.12.01

Coveralls badge, 100% test coverage via jimhester/covr.

2015.10.20

str_match_named was sometimes incorrectly issuing a warning when given
NA subjects, but this is now fixed.

When type.list is supplied, data.frame(stringsAsFactors=FALSE) is now
used to create the data.frame from the character match groups
matrix. This allows simple numerical conversions using functions like
as.numeric and as.integer (before, we got unexpected results since
those functions take the factor ID number).

2015.10.15

Test optional groups, set NA before calling substring to avoid "NAs
introduced by coercion" warning.

2015.10.07

stop with informative errors for pathological subjects and patterns.

2015.10.05

first version forked from str_match*perl in my .Rprofile. There are
several differences: 
- the "name" group is used for rownames, 
- the entire match is not returned (only the groups),
- str_match_named is vectorized (str_match_perl was not),
- names are taken from subject if present.
