[Ecls-list] About sockets - maybe last iteration?

[Matthew Mondor]

On Wed, 17 Feb 2010 14:54:13 +0100
Juan Jose Garcia-Ripoll <juanjose.garciaripoll at googlemail.com> wrote:

> I have revisited the SB-SOCKETS code and implemented the following changes.
[...]

Thanks for taking more time to work on this;  I just tested the changes
now and the socket does close when the stream closes as expected.  The
buffering being full by default is also nice, when I initially proposed
this it wasn't done in the official source because slime assumed an
unbuffered stream back then, if slime now also works with buffered
streams this is also great to know.

Perhaps a small issue is that I seem to see some unexpected (perhaps
temporary debug?) output at the ECL REPL at every connection to my test
multithreaded server:

#<input stream "FD-STREAM">
#<output stream "FD-STREAM">
#<two-way stream 0bd4cbe0>

I also saw about a twofold performance drop compared to with ECL 9, but
it well could be because of this extra output.

Thanks,
-- 
Matt