This choice is about how we should name the calendar classes.
The following class names are believed to be agreed:
These names are a gamble, based on how rapidly JSR-310 replaces the older API. I believe that it won't be that rapid, and as such the confusion simply isn't worth it.
This works by adding the least possible to get a unique class name. It has the advantage that IDE auto complete for a date or time will rapidly find the relevant class (although DateYMD is after DateTime alphabetically). The downside is that it uses abbreviations, and some may assume that only the year-month-day fields are accessible, and not day of week/week of weekyear etc.
Variations on the suffix theme might include Date310/Time310 or DateJSR/TimeJSR.
A 'calendar date' makes sense, but a calendar time' or 'calendar date and time' doesn't really. A desk calendar has dates on it, not times.
A 'wall time' makes sense (the time shown on a wall clock), but a 'wall date' or 'wall date and time' is stretching a little (not as bad as 'calendar time' though, as you could have a calendar on the wall).
Each element makes sense and is well named individually. But it is unclear that there is a connection between the three classes. ie. that CalendarDate + WallTime -> DateTime.
The no zone prefix refers to the lack of timezone. Describing a class by what it doesn't have seems backwards however.
The local prefix refers to the date/time/date-time being independent of timezone and local to the observer. It meets the common prefix test for easy learning - LocalDate + LocalTime -> LocalDateTime. Local is also a short prefix (being short does help on readability. Finally, it will be familiar to Joda-Time upgraders.
Apparantly, RFC2445 (icalendar) uses the 'local' terminology for "A local-time event at 5pm occurs at 5pm no matter what time zone the participant is in" (via Neal Gafter).