Philippe Marschall-2 wrote: > > On 10/11/2011 01:59 PM, Herby Vojčík wrote: >>> Without async IO supporting WebSockets does not seem to make much sense. >>> >>> Cheers >>> Philippe >>> >> >> I do not understand the rationale... "async IO" (more correctly, as Ryan >> Dahl often points, is "non-blocking IO", there is nothing async in >> node.js, >> it's single-threaded) is just the alternative model to "threaded, >> blocking >> IO". What does it have in common with sense of making WebSockets work? > > Pharo only supports "threaded, blocking IO", so for each WebSocket (or > long-polling, or forever i-frame) connection we need to have a Process > lingering around. That makes Pharo less than ideal for building a > socket.io server. > > Cheers > Philippe >
Thanx. That also answers the question if there is a project/plan for a socket.io server-side library (not, I assume) which I could reuse. So either write one (with knowing there will be lots of Processes involved), or not use Pharo for the server. -- View this message in context: http://forum.world.st/socket-io-tp3891592p3894239.html Sent from the Pharo Smalltalk mailing list archive at Nabble.com.
