Re: Patch: close I/O channels when a process ends

[David Daney <ddaney at avtrex dot com>]

David Daney wrote:
> Andrew Haley wrote:
> 
>>I ran out of file handles.  Was there any reason we weren't closing the
>>files in Process.waitFor?
>>
>>Andrew.
>>
>>
>>2004-08-04  Andrew Haley  <aph@redhat.com>
>>
>>	* java/lang/natPosixProcess.cc (waitFor): Close I/O channels.
>>
>>Index: java/lang/natPosixProcess.cc
>>===================================================================
>>RCS file: /cvs/gcc/gcc/libjava/java/lang/natPosixProcess.cc,v
>>retrieving revision 1.18
>>diff -c -2 -p -r1.18 natPosixProcess.cc
>>*** java/lang/natPosixProcess.cc	1 Mar 2004 21:33:27 -0000	1.18
>>--- java/lang/natPosixProcess.cc	4 Aug 2004 17:16:15 -0000
>>*************** java::lang::ConcreteProcess::waitFor (vo
>>*** 78,81 ****
>>--- 78,105 ----
>>      }
>>  
>>+   try
>>+     {
>>+       inputStream->close ();
>>+     }
>>+   catch (java::io::IOException *e)
>>+     {
>>+     }
>>+ 
>>+   try
>>+     {
>>+       outputStream->close ();
>>+     }
>>+   catch (java::io::IOException *e)
>>+     {
>>+     }
>>+ 
>>+   try
>>+     {
>>+       errorStream->close ();
>>+     }
>>+   catch (java::io::IOException *e)
>>+     {
>>+     }
>>+ 
>>    return status;
>>  }
> 
> 
> Why?
> 
I take it back.  But, what about any data in the pipes that is unread
when you do waitFor()?  It will now be lost.

David Daney