>> It's not entirely clear to me what the return value of File#delete() means.
> Did an error occur, or did the file simply not exist to begin with? If the
> latter, I think the code is probably fine as-is. If the former, we should
> probably check the return value and throw a QueryException if false.
>> 
> Note that file#delete will not throw an error if the delete is not possible,
> take a look here for example:
> http://www.exampledepot.com/egs/java.io/deletefile.html

Yes, I understand how to process a boolean return value.  :-)  But the Javadoc 
for File#delete() does not make it clear what the return value means (see 
above).

> So wouldn't be good to have the doDelete (indicating if the resource has
> been deleted or not, like in the doPut) return a boolean:
>            created = doPut(path, value);
> so we could have
>            deleted = doDelete(path, value);
> and behave like in the other method ...
> but maybe for the 2.1 to not break here method signature, ok ?

I don't see any need to change the method signature. If a caller tries to 
delete a file that doesn't exist, we return an HTTP 404. Returning a boolean in 
this case would be inconsistent with the HTTP spec.

>>> And last:
>>> tests/src/org/apache/pivot/tests/ShutdownTest.java:55
>>> Unchecked/unconfirmed
>>> cast from org.apache.pivot.wtk.Dialog to org.apache.pivot.wtk.Alert
>>> ...
>> Ignore all these. The casts are valid.
> Ok, but I guess if for the 2.1 could make sense to change a little their
> generics signature to have compile-time checks for these like <Y extends X>,
> but it's a small thing ...

They aren't generics.


Reply via email to