Re: Precise types & retyping
Daniel Bonniot <[email protected]>
| Newsgroups | gmane.comp.lang.nice.general |
|---|---|
| Message-ID | <[email protected]> |
Isaac Gouy wrote: >One of the benefits of Nice is that we can reuse existing Java >libraries. The tedious aspects of this are null testing or writing >retypings. > > The current policy (in the absence of retyping) is to choose the most convenient nullness for Java types (that is, parameters can be null, but results are guarenteed not to be null). This means that the compiler should not bother you, so that should not be tedious. Of course, you don't get the extra safety you get when using Nice libraries, but there is no simple automatic way to do so. You can get the safety back by writing the retypings. There is a plan to allow code to set a different retyping policy for a whole class or a whole package. That should help limit the number of necessary retypings by picking the right policy, and then enumerating the few special cases. >Is there a way to deal with multiple Java class constructors in a >retyping? > > You mean retype several constructors in one go? There will be the way mentioned above. What else can you imagine? >Which methods/fields/classes/interfaces... actually need retyping? > > None, until you find a reason to add a retyping ;-) Seriously, you should not need to write retypings in most cases, so wait to find a need for them. One main reason is a type that should be parameterized, like List, since Nice can not guess what kind of List it is, so you have to tell it. >Would it be possible to use reflection to automatically write Nice >retypings for a Java library? > > How would it guess the type information that is not there (nullness, parameterized types)? Daniel ------------------------------------------------------- SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now! http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click