Re: Re: [PEPr] Comment on HTTP::HTTP_SessionServer

From: Date: Sun, 17 Oct 2004 12:09:50 +0000
Subject: Re: Re: [PEPr] Comment on HTTP::HTTP_SessionServer
References: 1 2  Groups: php.pear.dev 
Request: Send a blank email to pear-dev+get-33905@lists.php.net to get a copy of this message
Hi, Stephan Schmidt wrote:
Olivier Guilyardi wrote:
The protocol you use looks simple and efficient. But wouldn't it be more portable to write a msession client in PHP ? It could be loaded in case the msession extension is not available.
This package is not about the client, but a server implementation in PHP. Writing a client for msession shouldn't be very hard, but this should be Net_MSession and would only be using fsockopen().
Okay, sorry for this confusion. But since you provide both client and server, this was not clear to me. And actually, Actually, what's behind my comment is the following argument : you state that HTTP_SessionServer allows to share data between applications and/or programming languages, but you create a new protocol and API. My point is : wouldn't it be more efficient to support an already existing protocol (as msession) so that one could make use of all of the existing API, wether they be C, Java, etc... ? I have another question : Say that I want to do some poor-man load-balancing : 1 - I have several cheap/free accounts at different hosting providers 2 - People who access the primary server then get redirected to one of the other
    ones, according to some simple load/ping statistics
3 - I need to share session data but I can't run a standalone server on a port
    I choose on any of these basic http accounts.
4 - In this situation, I could use of one my server as a session server.
    But it would require the protocol to be wrapped into HTTP requests (XML-RPC,
    etc...).
What do you think about this ? Have you been thinking of such HTTP-encapsulation ? -- og

« previous php.pear.dev (#33905) next »