Well, my 'try' was successful. AFAIK, it was the first full non-blocking ssl filter. We have yet to have the changes we made make it into POE core though :)
I'm glad you're spear heading this, and Data::Transform is a good choice. I'm ok with a patch to Wheel::ReadWrite as long as it is backward compatible and the performance penalty is minimal. What does everyone else think? David Davis ☄ Software Engineer http://xant.us/ On Sat, Jul 19, 2008 at 10:38, Martijn van Beers <[EMAIL PROTECTED]> wrote: > Hi, > > I've been working on a filter to do SSL, inspired by David Davis' try in > Sprocket and IO::Socket::SSL. It seems to be working pretty well (see > attached code; mostly needs more testing and error handling), but I > can't release it because it needs more API than filters currently > provide (see also the 'POE::Filter error reporting' thread) > > My current thoughts are to forget about trying to extend POE::Filter but > start over in a new namespace (possibly Data::Transform?) That way > there's no need to worry about mixing up filters with the old and new > apis. And not being in the POE namespace might help convince people that > Filters really are useful outside of POE. > > Of course, there is still need for handling this in Wheel::ReadWrite > (and other wheels) too. Not sure whether to subclass those or to add > some extra if blocks. > > I'm mostly writing this email to find out what people think of the idea > of doing 'new filters' in a new namespace (and if so, which namespace). > But if you think I'm on crack for how I'm doing things in Filter::SSL, > please let me know that too :) > > > Martijn >