> It is not possible to set a zero second timeout for accepts() in order to do > a pure poll on a IO::Socket. Should the code checking to see if a timeout > has been set be: (sorry for the long delay in answering) Well, I now added the capability to do a "poll" with the accept() and connect() -- but I'm doing it only in the spirit of TIMTOWTDI, the principle of least surpise, and consistency with select(), because you really shouldn't be doing that: +The timeout in the [PKG] can be specified as zero to effect a "poll", +but you shouldn't do that because a new IO::Select object will be +created behind the scenes just do to the single poll. This is +horrendously inefficient. Use rather true select() with a zero +timeout on the handle, or non-blocking IO. -- $jhi++; # http://www.iki.fi/jhi/ # There is this special biologist word we use for 'stable'. # It is 'dead'. -- Jack CohenThread Previous | Thread Next