Thank's Alex, thats extremly useful.
I think there are ways and means to sort out state tracking in an asyncronous server. If your familer with Python I'd be going for something along the lines of what the Twisted framework and its differed execution model. My other critisism of the server in lib/http.l is that it does not sufficently abstract the URL space from the physical disk. Most other mature frameworks allow your to creat a completly virtual url space which has no relation to physical files on disk. The way the forms.l maps a full path to its javascript file and puts it in the output makes me particularly uncomfortable as it is exposing data on where I have installed picolisp. I'm also a big fan of having human readable urls which do not include any cryptic characters like @ signs etc. I'll expriment with some url spaces. My first inclination would be to setup a simple tree which can be traversed based in the url broken up on '/' delimiters. Any delimiters left over become arguments to whatever function gets called. regs Konrad. 2008/10/7 Alexander Burger <[EMAIL PROTECTED]>: > Hi Konrad, > >> Are their any existing picoLisp functions for using the select system >> call and doing asynchronous IO > > Yes. In fact, this is the basis of almost all activities in a typical > application. > > The main control structure is a list of file descriptors and expressions > in the global '*Run', usually in combination with the interface function > 'task'. > > > For example, if you want to listen on multiple ports: > > (for P (4001 4002 4003 4004) > (task (port P) > (when (accept @) > (task @ > Sock @ > (in Sock > (if (rd) > (out Sock (eval @)) > (task Sock) > (close Sock) ) ) ) ) ) ) > > This will listen on those four ports, and whenever a connection request > arrives, it will install a new task on this socket. Then, whenever data > arrive on that socket, they are taken as an 'exe' expression, and get > evaluated with the current output channel set to that socket. If EOF > arrives instead, the connection is closed and the task un-installed. > > This happens all in the "background". You may still have some main loop > in your program (or, if not, you will be dropped into the console ':' > prompt). The "do nothing" main loop, BTW, is (wait), that's why the > examples for applications often have a '-wait' on the command line. > > > There are endless ways to combine 'task' with I/O functions. A simple > "single shot" background task, accepting a request on port 4444, reading > an expression, evaluating it and immediately closing the connection: > > (task (port 4444) > (let? Sock (accept @) > (in Sock (out Sock (eval (rd)))) > (close Sock) ) ) > > Or, a main loop keeping multiple connections open on port 4444: > > (setq P (port 4444)) > (loop > (task (listen P) > (in (setq Sock @) > (if (rd) > (out Sock (eval @)) > (task Sock) > (close Sock) ) ) ) ) > > > >> for anyone curious I find myself desiring an asyncronous http server. >> rather than the current forking server. This I know is not currently > > I preferred a forking server simply because it makes it easier to keep > the state and context for each client. > > >> present (in fact the server in http.l seems to be very strongly tied >> to the gui framework, and hence makes a lot of assumptions that I want > > I do not feel that "lib/http.l" is tied to the 'xhtml' and 'form' parts. > Originally, I developed it for the Java GUI (the "lib/gui.l" and > "lib/gui2.l" files, no longer in the standard release). You just should > not use the 'server' function from "lib/http.l", but write something > along the lines of the above examples. > > Cheers, > - Alex > -- > UNSUBSCRIBE: mailto:[EMAIL PROTECTED] > -- UNSUBSCRIBE: mailto:[EMAIL PROTECTED]
