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

[Andrew Haley <aph at redhat dot com>]

David Daney writes:
 > Andrew Haley wrote:
 > > David Daney writes:
 > >  > Tom Tromey wrote:
 > >  > >>>>>>"Andrew" == Andrew Haley <aph@redhat.com> writes:
 > >  > > 
 > >  > > 
 > >  > > Andrew> I ran out of file handles.  Was there any reason we weren't
 > >  > > Andrew> closing the files in Process.waitFor?
 > >  > > 
 > >  > > It probably isn't valid to do this.
 > >  > > Couldn't there still be data in the OS' pipe buffer?
 > >  > > 
 > >  > > Tom
 > >  > 
 > >  > My patch that is waiting for approval is an almost total rewrite of
 > >  > natPosixProcess, so would replace this.  However it would not solve the
 > >  > problem either.
 > >  > 
 > >  > I think that the place to fix this is in
 > >  > FileInputStream/FileChannelImpl.  When you get end-of-file when reading
 > >  > a pipe/file, you should close the descriptor there.
 > > 
 > > But we can still close the output channels here, can we?
 > > 
 > 
 > The only thing that I think is safe would be the following:
 > 
 > For each of inputStream and errorStream, when the Process terminates if
 > the corresponding get[InputStream|ErrorStream]() has not been called AND
 >  there are no bytes available, we can safely assume that no more data
 > will be coming.  Only if these conditions are met can you close the
 > streams.  However after closing the streams, you must replace them with
 > dummy streams that always return EOF.

Seems like it would work.

 > Likewise for the outputStream, if getOutputStream() has not been called,
 > you can close the stream and replace it with a dummy stream that always
 > throws IOException.
 > 
 > If this sounds reasonable, I can integrate this into my PosixProcess
 > patch which can be found at:
 > 
 > http://gcc.gnu.org/ml/java-patches/2004-q3/msg00452.html

I'm in favour of anything that will fix the bug, I guess.

 > Opinions?

Why not read the stream till EOF, buffer it, and close the underlying
file handles?  They can't be used anyway because the process they're
connected to has died.

Andrew.