Re: natPlainSocketImplWin32 fix and natPlainSocketImplPosix adjustment

[David Daney <ddaney at avtrex dot com>]

Marco Trudel wrote:
> Hello list
>
> If a remote host closes a socket connection, posix throws an 
> IOException. Win32 doesn't do anything because WSAECONNRESET is set 
> and this is ignored (it looks like it's the equivalent to the "Broken 
> pipe" errno on posix).
> The consequences are, that on win32 an application just continues 
> sending data until all data is sent.
>
> Another thing is, that sun jvms throw SocketException, not 
> IOException. I adapted both files for that.
>
> The copyright assignment isn't done yet, but I think we can discuss 
> the patch anyway. I will repost it (as well as the security patch) as 
> soon as the assignment stuff has taken place.
>
> Comments, hints?
I don't know about win32, but fixing the exception type is the proper 
thing to do.  Typically for problems like this a test case would be 
added to mauve.

The legal requirements for committing to mauve are quite a bit less 
rigorous than for classpath and gcc.  Perhaps someone (Tom Tromey?) 
could give you commit access for mauve.

David Daney