2008/10/2 Steve Ulrich <[EMAIL PROTECTED]>:
> Hi!
>
>> Niklas Gustavsson [mailto:[EMAIL PROTECTED] wrote
>>
>> Yes, I'm aware that we can chain and replace exceptions thrown by the
>> various protocols, that was not my concern. My question was if
>> IOExceptions semantics matches the error conditions that we talk about
>> here. From the Javadoc: "Signals that an I/O exception of some sort
>> has occurred". I'm not sure if this is true for many of the cases
>> where the FtpServer file system abstraction fails. But then again,
>> this might be a case of splitting hairs. What do the rest of you
>> think?
>
>
> Using IOException for anything not IO related isn't a good practice.
I/O mean input/oputput, it does not cover only the I/O to real
filesystem it cover also I/O trough the net sockets, the
RemoteException used in RMI is an IOException.

An error that occour while accessing to a virtual file system for me
is equal to an error that occour in a real filesystem.
In the standard java library all error that occour accessing to a real
filesystem are IOException.

> If there's really an IOException the implementation must cleanup the IO 
> resources!
Which implementation? the FileSystemView implementation or the
FileSystemView client code?

> Misusing IOException will lead to confusion and zombie connections/IO 
> resources.
I'm not conviced, how this should happen (in the given case)?


-- 
Andrea Francia
http://andreafrancia.blogspot.com/

Reply via email to