Re: Changes to java/net/PlainSocketImpl.java

[Michael Koch]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Am Mittwoch, 30. Juli 2003 15:31 schrieb Stephen Crawley:
> Hi Torsten,
>
> I don't know if you did this deliberately, but your latest change
> to java/net/PlainSocketImpl.java to implement the
> connect(addr,timeout) method has clobbered an existing
> implementation checked in a few hours earlier by Michael Koch on my
> behalf.
>
> Your implementation ignores the timeout (with a comment to the
> effect that this should be implemented in the future)
>
> The implementation that yours overwrote implement connect timeouts
> by setting the SO_TIMEOUT socket option before calling the native
> connect method.
>
> I know that my implementation will not work for older Linux kernel
> that do not allow you to change socket timeouts. But recent kernels
> do ... by to my reading of the kernel source code. And at least the
> behaviour in an older kernel will be to use a default timeout ...
> which is the same as what your implementation does!
>
> The alternative approach to implementing the connect timeout would
> entail setting an interval timer and catching SIGALRM, etc.  I
> don't think we should do this in the Classpath native libs because
> of the possibility of interfering with other uses of SIGALRM by the
> JVM.
>
> Either way, do you intend to implement connect timeouts?

I reverted Torstens commit. Stephens implementation is better and more 
complete. I think it was an accidental commit by Torsten.

Torsten: CVS has surely gicen you a merging conflict before commit. 
Just copying over your working version of the file is __NNOO__ 
solution. Its evil.


Michael
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQE/J9mrWSOgCCdjSDsRAi+YAJ9yJ6313eG5r7SHqwtYm7MJSPKL7ACeJemH
8bnI2rztsohrjRWKx9JOx04=
=tly4
-----END PGP SIGNATURE-----