I have some incomplete stream code that fixed other issues, which I believe are unrelated to what you mention. From what I recall, non-blocking streams in iolib block under certain circumstances. My code added an additional buffer mechanism to deal with this issue, among other things.
I believe my clone is on github, but i might have lost the most important changes in a freak accident. Best, Red On Thu, Oct 7, 2010 at 6:17 AM, Attila Lendvai <[email protected]> wrote: > dead list, > > i've recently encountered a headache: a machine was connecting to our > iolib based web server, and hanged all the worker threads. the problem > is that the connection went try, but it didn't get disconnected after > several minutes. > > the hang is somewhere inside (defmethod stream-read-byte ((stream > dual-channel-gray-stream)) ...). > > the proper fix would be to implement the call/cc based connection > multiplexing and gc dry/slow connections. i'm going to work on it in > the upcoming weeks, but i wonder what are the plans regarding the > upcoming iolib development? > > Stelian said that he has plans with the streams, notably to prefer > zstreams eventually, but it's not ready yet. > > anything else? > > -- > attila > > _______________________________________________ > IOLib-devel mailing list > [email protected] > http://common-lisp.net/cgi-bin/mailman/listinfo/iolib-devel > _______________________________________________ IOLib-devel mailing list [email protected] http://common-lisp.net/cgi-bin/mailman/listinfo/iolib-devel
